TWI363963B - Automated abnormal machine tracking and notifying system and method - Google Patents

Automated abnormal machine tracking and notifying system and method Download PDF

Info

Publication number
TWI363963B
TWI363963B TW96150638A TW96150638A TWI363963B TW I363963 B TWI363963 B TW I363963B TW 96150638 A TW96150638 A TW 96150638A TW 96150638 A TW96150638 A TW 96150638A TW I363963 B TWI363963 B TW I363963B
Authority
TW
Taiwan
Prior art keywords
machine
abnormal
information
abnormality
user
Prior art date
Application number
TW96150638A
Other languages
Chinese (zh)
Other versions
TW200928726A (en
Inventor
Tai Chuan Chiang
Chan Pai Liu
Original Assignee
United Microelectronics 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 United Microelectronics Corp filed Critical United Microelectronics Corp
Priority to TW96150638A priority Critical patent/TWI363963B/en
Publication of TW200928726A publication Critical patent/TW200928726A/en
Application granted granted Critical
Publication of TWI363963B publication Critical patent/TWI363963B/en

Links

Landscapes

  • General Factory Administration (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Description

!363963 . 九、發明說明: 【發明所屬之技術領域】 本發明係關於-種機台異常自動化追蹤及通報系統及 方法,尤指一種可整合不同異常資訊並即時反映可能異常 機台之自動化追蹤及通報系統及方法。 【先前技術】 _ 現今的製造產業,尤其是半導體晶圓廠或是液晶顯示面 板製造廠等,涉及非常複雜且繁多的製造流程步驟。一個 產品在製作過程中必須行經不同的製造機台、傳送搬運機 台、儲存機台,並歷經數十道甚或數百道的製程才完成。 而各道製程在量產真的產品或試產不同的技術或產品前, 須經過多次的試驗、參數微調,以達到特定的生產需求, 同時確保及控管各段製程及產出產品的品質。 ® —般在生產線品管部門的工作,係針對一特定製程產出 之特定產品進行測試,並由工程師針對發生異常的管制異 常批(out of control,OOC lot)或暫停批(hold lot)的統計製程 控制報表(statistical process contro卜 SPC chart)或產品晶圓 允收測試(wafer acceptance test,WAT)數據等進行分析,尋 找出異常產品與異常機台之關聯性,來指認並診斷出可能 的異常機台。但此一以產品或製程為對象的分析方法,常 會忽略某些異常機台在未造成明顯影響前的變異。因此往 6 1363963 往不能在下一批次(next lot)進入可能的異常機台前便及時 發現機台的異常狀態,而造成產品良率與生產成本的損失。 另外,由於生產線上使用的生產、檢測工具甚多,不同 的操作人員通常僅負責部分的生產、檢測工具,且操作人 員多半係以輪班方式進行品管工作。雖然各製造廠内莫不 透過一文管或電子交接系統進行交班,但由於對人為分析 結果的標準判斷不一,又或者有遺漏交接事項等狀況的發 生,往往使得被誤判的單一事件或微小事件引發或擴大成 後續重大的異常,造成產品良率與生產成本的損失。 由此可知,如何能跨越各班次而自動追蹤及整合分析各 製程機台不同的異常資訊,進而自動化地追蹤並通報可能 異常的機台,以提高生產製程的穩定度,實為一值得研究 之議題。 【發明内容】 因此,本發明之一目的係在於提供一種可跨越各班次 自動整合各製程機台的異常資訊,並歸納出可能異常的機 台,進而自動追蹤與通報之系統與方法。 根據本發明所提供之申請專利範圍’係提供一種機台 異常之自動化追蹤及通報系統。該系統係包含:一用以自 1363963 動擷取複數個檢測結果,並分別整合成複數個異常資訊之 資料擷取與分析子系統;一用以儲存該等異常資訊之機台 狀態資料庫;一用以根據該等異常資訊與至少一相關之可 能異常機台來分別計算個別可能異常機台之異常得分之異 常資訊計算裝置;一用以判斷個別可能異常機台之一異常 得分總和是否大於η值之判斷裝置,而當該異常得分總 和大於該η值時,即判斷須通知至少一使用者。該系統更 包含一用以將該可能異常機台與相關之該等異常資訊通報 予該使用者之自動通報子系統。 根據本發明之申請專利範圍,更提供一種機台異常自動 化追蹤及通報之方法。該方法包含有以下步驟··首先利用 一資料擷取與分析子系統自動擷取複數個檢測結果並分別 整合分析成複數個異常資訊,而該等異常資訊係儲存於一 機台狀態資料庫。接下來,利用一異常資訊計算裝置根據 該等異常資訊與至少一相關之可能異常機台來分別計算個 別異常機台的異常得分,並利用一判斷裝置判斷個別可能 異常機台之一異常得分總和是否大於一 η值。當該總和大 於η值時,即利用一自動通報子系統將該可能異常機台與 相關之該等異常資訊通報予至少一使用者。 根據本發明所提供之機台異常之自動化追蹤及通報之 系統與方法,異常資訊係藉由該資料擷取與分析子系統自 8 1363963363963. IX. Description of the Invention: [Technical Field] The present invention relates to a system and method for abnormal automatic tracking and notification of a machine, in particular, an automatic tracking that can integrate different abnormal information and instantly reflect possible abnormal machines. And notification systems and methods. [Prior Art] _ Today's manufacturing industry, especially semiconductor fabs or LCD panel manufacturers, involves very complex and numerous manufacturing process steps. A product must be manufactured through different manufacturing machines, transfer handlers, storage machines, and after dozens or even hundreds of processes. Each process must undergo repeated tests and fine-tuning of parameters to achieve specific production requirements, while ensuring and controlling the various processes and products produced before mass production of real products or trial production of different technologies or products. quality. ® generally works in the quality control department of the production line, testing specific products produced by a specific process, and is engineered to out of control (OOC lot) or hold lot. Statistical process control report (statistical process contro SPC chart) or product wafer acceptance test (wafer acceptance test (WAT) data, etc., to find out the relationship between abnormal products and abnormal machines, to identify and diagnose possible Abnormal machine. However, this analysis method that targets products or processes often ignores the variation of some abnormal machines before they cause significant impact. Therefore, to 6 1363963, the abnormal state of the machine can not be found in time before the next lot enters the possible abnormal machine, resulting in loss of product yield and production cost. In addition, due to the large number of production and inspection tools used on the production line, different operators are usually only responsible for part of the production and testing tools, and most of the operators are performing quality control work in shifts. Although the manufacturing companies do not pass through a document management or electronic handover system, due to the different judgments on the results of human analysis, or the occurrence of missing handovers, etc., the single event or minor event that is misjudged is often triggered. Or expand into subsequent major anomalies, resulting in loss of product yield and production costs. It can be seen that how to automatically track and integrate different abnormal information of each process machine across each shift, and then automatically track and report the abnormal machine to improve the stability of the production process, it is worth studying. The issue. SUMMARY OF THE INVENTION Accordingly, it is an object of the present invention to provide a system and method for automatically tracking and reporting anomaly information that automatically integrates various process machines across shifts and that summarizes possible abnormalities. The patent application scope provided in accordance with the present invention provides an automated tracking and notification system for machine anomalies. The system comprises: a data acquisition and analysis subsystem for extracting a plurality of test results from 1363963 and integrating into a plurality of abnormal information; a machine state database for storing the abnormal information; An abnormality information computing device for respectively calculating an abnormality score of an individual possible abnormal machine according to the abnormal information and at least one related abnormal machine; and determining whether the sum of the abnormal scores of one of the individual abnormal machines is greater than The η value judging means, and when the sum of the abnormality scores is greater than the η value, it is judged that at least one user is to be notified. The system further includes an automatic notification subsystem for communicating the possible abnormal machine and associated abnormal information to the user. According to the patent application scope of the present invention, a method for automatically tracking and notifying the abnormality of the machine is provided. The method includes the following steps: First, a data acquisition and analysis subsystem automatically extracts a plurality of detection results and integrates and analyzes the plurality of abnormalities into a plurality of abnormal information, and the abnormal information is stored in a machine state database. Next, an abnormality information computing device is used to calculate the abnormality scores of the individual abnormal machines according to the abnormal information and the at least one related abnormal machine, and use a determining device to determine the sum of the abnormal scores of one of the individual abnormal machines. Whether it is greater than an η value. When the sum is greater than the value of η, an automatic notification subsystem is used to notify the at least one user of the possible abnormal machine and related abnormal information. The system and method for automatic tracking and notification of machine abnormalities according to the present invention, the abnormal information is obtained by the data extraction and analysis subsystem from 8 1363963

• 動揭取線上、或離線之各階段檢測結果或使用之輸入資 料’並藉由該分析與判斷裝置分析整合該等檢測結果與輸 入資料所得’並將該等異常資訊儲存於該機台狀態資料庫 中。而當該等異常資訊計算之得分總和大於該η值時,則 將該可能異常機台與相關之該等異常資訊自動通報予使用 者。由此可知本發明所提供之機台異常之自動化追蹤及通 報之方法可自動追蹤、統計以及分析異常資訊與相關機台 • 的關聯性’而歸納診斷出異常機台或可能異常之機台,並 即時通報使用者。且由於上述方法中的各步驟皆係利用本 發明所提供之異常機台之自動化追蹤及通報系統所完成, 因此可省去人工操作之影響與彌補使用者經驗之不足’故 本發明更為一可跨越不同班次、不同製程、與不同機台而 提供完整的異常資訊與異常機台相關性之系統與方法Q 【實施方式】 鲁請參閱第1圖,帛1圖係為本發明所提供之機台異常之 自動化追蹤及通報系統之一較佳實施例示意圖。如第i圖 所示,本發明所提供之機台異常之自動化追蹤及通報系統 100包含有一資料擷取與分析子系統102、一載入裝置 104、一機台狀態-貝料庫1 〇6、一異常資訊計算裝置1 〇8、 一判斷裝置110、以及一自動通報子系統112。資料擷取與 分析子系統102係用以自動擷取複數個檢測結果,例如線 上(on-line)製程數據、離線(0ff_〗ine)製程數據、機台監測 1363963 (machine monitor)數據、產品缺陷(product defect)數據、或 使用者之輸入資料如顯示保養時間點、更換零件、檢閱結 果之註解(comment)等,將該等檢測結果或輸入資料分析整 合成複數個異常資訊,並藉由載入裝置104將該等異常資 訊自動載入儲存一機台狀態資料庫106。儲存於機台狀態 資料庫106之該等異常資訊至少包含有日期、產品批次、 缺陷種類、缺陷程度、相關可能異常機台、與該可能異常 機台之異常得分等資料。此外,資料擷取與分析子系統102 亦可用以根據該異常資訊分析及診斷出至少一可能之異常 機台,並將該機台之異常資訊自動載入機台狀態資料庫106 中。載入裝置104更可用以人工載入該異常資訊’或人工 載入任一非來自於資料棟取及分析子系統102之新異常資 訊。 異常資訊計算裝置108則用以根據該等異常資訊與至 * 少一相關之可能異常機台來分別計算個別可能異常機台之 異常得分;而判斷裝置110則用以判斷個別可能異常機台 之一異常得分總和是否大於一 η值。當該異常得分總和大 於η值時,例如,當η值大於等於(g ) 2,判斷裝置110即 判斷該可能異常機台與相關之該等異常資訊須通報予至少 一使用者。值得注意的是,判斷裝置11〇係用以判斷一預 定週期内之該等異常得分總和,當該預定週期内之該異常 • 得分總和大於該η值時,始判斷須將該可能異常機台與相 1363963 關之該等異常資訊通報予該使用者。該n值係可根據產業 類別、製程複雜程度、機台種類、機台產出量、機台停機 時間、或機台預防性保養循環(preventive maintenance,ΡΜ cycle)而由使用者定義或調整。最後自動通報子系統112係 用以藉由電子信件、電腦整合製造(computer integrated manufacture,CIM)系統之警示裝置、手機簡訊、機台操作 介面警示訊號、或終端使用者電腦螢幕等途徑將該可能異 φ 常機台與相關之該等異常資訊通報予使用者。 請同時參閱第1圖與第2圖,其中第2圖係為本發明所 提供之機台異常之自動化追蹤及通報之方法之一第一流程 示意圖。一般而言,使用者如線上製程工程師或品管工程 師會針對不同製程之不同產品進行檢測,例如製程驗證 (process qualification)相關測試、產品驗證(product qualification)相關測试、製程評估(pr〇cess evaiuati〇n)相關 測試、製程監測(process monitor)、定期產品監測(routine product monitor)、在線製程評估與監測(in_Hne process evaluation & monitor)等,而獲得線上製程數據、離線製程 數據、機台k測數據、或產品缺陷數據等。此一大量的測 試與監測結果數據即所謂的原始資料,線上製程工程師或 品管工程師會根據上述原始資料判別某一批次之產品是否 異常。如第2圖所示’經判定為報廢批(scrapl〇〇、管制異 常批(OOC lot)或暫停批(h〇ld 1〇t)的統計製程控制報表(spc 1363963 •曰rt)或產品晶圓允收測試(wat)數據等則再經由工八 析得到-檢測結果,以尋找出異常產品與異 二 性’來指認並診斷出至少一可能的異常機台,且工程= 將此一異常資訊或直接將檢測結果輸人—註解(咖啊/ 中。接下來,如第2圖所示,本發明所提供之機台 自動化追蹤及通報之方法包含有以下步驟:• Uncovering the results of the various stages of the online or offline, or using the input data 'and analyzing and integrating the test results and input data by the analysis and judgment device' and storing the abnormal information in the state of the machine In the database. When the sum of the scores calculated by the abnormal information is greater than the value of η, the possible abnormal machine and the related abnormal information are automatically notified to the user. Therefore, it can be seen that the method for automatic tracking and notification of the abnormality of the machine provided by the present invention can automatically track, statistically analyze and analyze the relationship between the abnormal information and the related machine, and summarize the machine that diagnoses the abnormal machine or the abnormal machine. And notify the user immediately. Moreover, since each step in the above method is completed by using the automatic tracking and notification system of the abnormal machine provided by the present invention, the influence of the manual operation and the deficiency of the user experience can be saved, so the present invention is more System and method for providing complete abnormal information and abnormal machine correlation across different shifts, different processes, and different machines. [Embodiment] Please refer to Figure 1, Figure 1 is provided for the present invention. A schematic diagram of one preferred embodiment of an automated tracking and notification system for machine anomalies. As shown in the figure i, the automatic tracking and notification system 100 for the machine abnormality provided by the present invention comprises a data acquisition and analysis subsystem 102, a loading device 104, and a machine state-before the library 1 〇 6 An abnormality information computing device 1 〇8, a determining device 110, and an automatic notification subsystem 112. The data acquisition and analysis subsystem 102 is configured to automatically retrieve a plurality of detection results, such as on-line process data, offline (0ff_〗 〖) process data, machine monitoring 1363963 (machine monitor) data, product defects (product defect) data, or input data of the user, such as displaying maintenance time points, replacing parts, commenting on the review results, etc., and integrating the test results or input data analysis into a plurality of abnormal information, and The ingress device 104 automatically loads the exception information into the stored machine state database 106. The abnormal information stored in the machine state database 106 includes at least the date, the product batch, the defect type, the defect degree, the related abnormal machine, and the abnormality score of the possible abnormal machine. In addition, the data capture and analysis subsystem 102 can also be used to analyze and diagnose at least one possible abnormal machine based on the abnormal information, and automatically load the abnormal information of the machine into the machine state database 106. The loading device 104 can be used to manually load the exception information or manually load any new anomaly information that is not from the data acquisition and analysis subsystem 102. The abnormality information computing device 108 is configured to separately calculate an abnormality score of the individual possible abnormal machine according to the abnormal information associated with the abnormal information, and the determining device 110 is configured to determine the individual abnormal machine. Whether the sum of the abnormal scores is greater than an η value. When the sum of the abnormal scores is greater than the value of η, for example, when the value of η is greater than or equal to (g) 2, the determining means 110 determines that the abnormal machine and the associated abnormal information are to be notified to at least one user. It is to be noted that the judging device 11 is configured to determine the sum of the abnormal scores in a predetermined period. When the sum of the abnormalities and the scores in the predetermined period is greater than the η value, it is determined that the abnormal machine must be The abnormal information related to the phase 1363963 is notified to the user. The n value can be defined or adjusted by the user according to the industry category, process complexity, machine type, machine output, machine downtime, or machine preventive maintenance (ΡΜ cycle). Finally, the automatic notification subsystem 112 is used to enable the possibility by means of an electronic mail, a computer integrated manufacturing (CIM) system warning device, a mobile phone newsletter, a machine operation interface warning signal, or a terminal user computer screen. The abnormal φ machine and the related abnormal information are notified to the user. Please refer to Fig. 1 and Fig. 2 at the same time. Fig. 2 is a first flow diagram showing one of the methods for automatic tracking and notification of machine abnormalities provided by the present invention. In general, users such as online process engineers or quality control engineers will test different products for different processes, such as process qualification related tests, product qualification related tests, and process evaluations (pr〇cess). Evaiuati〇n) related test, process monitor, routine product monitor, in_Hne process evaluation and monitoring, etc., to obtain online process data, offline process data, machine k measurement data, or product defect data. This large amount of test and monitoring result data is the so-called original data, and the online process engineer or quality control engineer will judge whether a batch of products is abnormal based on the above-mentioned original data. As shown in Figure 2, the statistical process control report (spc 1363963 • 曰rt) or product crystal determined to be scrapped (scrapl〇〇, controlled abnormal batch (OOC lot) or suspended batch (h〇ld 1〇t) The round acceptance test (wat) data, etc., is then obtained through the analysis of the work - to find out the abnormal product and the heterosexuality' to identify and diagnose at least one possible abnormal machine, and the engineering = this abnormality Information or directly enter the test results - annotation (Caf / in. Next, as shown in Figure 2, the method for automated tracking and notification of the machine provided by the present invention comprises the following steps:

步驟200:本發明所提供之機台異常之自動化追縱及通報 系統100係利用資料擷取與分析子系統1〇2係自動擷取 工程師之輸入資料,例如註解内之該等異常資訊;資料 擷取與分析子系統亦可直接自動擷取上述檢測結果並整 合分析而得到複數個異常資訊。 步驟210:接續步驟200。隨後載入裝置104係用以將該等 異常資δίΐ自動載入並儲存於機台狀態資料庫中。值 得注意的是,該等異常資訊亦可藉由人工經由載入震置 104載入及儲存於機台狀態資料庫1〇6。另外,當—新機 台進入製程時,亦可透過載入裝置1〇4將新機台之相關 異中> βίΐ人工載入於機台狀態資料庫1〇6内。儲存於機 台狀態資料庫106之異常資訊則至少包含有日期、產品 批次、缺陷種類、缺陷程度、相關之可能異常機台、與 該可能異常機台之異常得分等資料。 步驟220:接續步驟210。利用異常資訊計算裝置1〇8根據 該異常資訊與至少一相關之可能異常機台分別計算個別 可月b異吊機台的異常付分。由於各機台產出(throughput) 12 曝63963 與停機時間(machine idle time)、預防性保養循環 (preventive maintenance,PM cycle)皆不同,因此異常資 訊計算裝置108在計算該等異常得分時,係計算一預定 週期内之異常得分;而週期外之異常得分則可忽略或不 列入計算,但仍可儲存於機台狀態資料庫106中作為之 後異常機台之協助診斷參考資料。 步·驟230 :接續步驟2加。利用判斷裝置110來判斷個別可 能異常機台之一異常得分總合是否大於一 η值。 梦驟240:接續步驟230。當該異常得分總和在該預定週期 内大於該η值時’判斷裝置11〇即判斷將該可能異常機 台與相關之該等異常資訊利用自動通報子系統112通報 予至少一使用者。而當該異常得分總和小於η值;甚至 在超過該預定週期之後仍小於η值時,則不列入下次預 定週期内之異常得分計算,但仍儲存於機台狀態資料庫 106中作為之後協助異常機台之診斷參考資料,直到使 用者認為該異常資訊及異常得分可歸零或刪除。 以下為本發明根據上述第一流程步驟所提供之第一較 佳實施例。當歷經第一機台完成多晶矽層微影(p〇ly i ph〇t〇) 製作之第一批次產品經檢測發現有管制異常(ooc)之狀況 發生’並被認定為暫停批並暫停下來,㈣經由卫程 認此第-批次產品發生曝光失焦(defocus)與掉落微粒 (dropped particle)等缺陷。若缺陷嚴重而達到報廢栌 13 1363963 第一批次即視為報廢批,同時第一機台即須立即調校。若 尚未達到報廢標準,且工程師判定可接續流程,則會在釋 出第一批次的同時輸入一註解,例如:第一批次產品有一 晶粒發生失焦,並有50%發生微粒掉落缺陷。 本發明所提供之機台異常之自動化追蹤及通報系統 100係如步驟200所述,在不同的製程結束後,即利用資 料擷取與分析子系統102立即自動擷取並分析工程師註解 内之該等異常資訊,故第一批次產品係得到如表格1所示 之異常資訊: 表格1 曰 期 批 次 缺陷 種類 發生 數量 發生範 圍(%) 致命缺陷/ 非致命缺 陷 機 台 機台 異常 得分 發現層次 層次 得分 第 失焦 1 N/a 致命 第 1 Poly 1 photo 1 第 一一 微粒 掉落 N/a 50 致命 第 一 1 Poly 1 photo 1 值得注意的是,資料擷取與分析子系統102對於工程 師未註解之部分可暫時以n/a表示,且會自動擷取該批次 剛經歷之機台,而於表格1之異常資訊内顯示為第一機 14 1363963 台。此外,使用者可預先設定不同製程與不同機台可能發 生的缺陷種類為致命缺陷(killer defect)或非致命缺陷 (non-killer defect),而資料擷取與分析子系統102則依照使 用者所預定之規則將註解内的缺陷自動分類,並記錄為致 命缺陷或非致命缺陷。 接下來,則如步驟210所述,本發明所提供之機台異 常之自動化追蹤及通報系統100係利用載入裝置104將表 格1所載之該等異常資訊儲存於機台狀態資料庫106中。 並如步驟220所述,利用異常資訊計算裝置108分別給予 該等異常資訊一異常得分,如表格1所標示在機台異常得 分一攔位予以一異常得分;異常資訊計算裝置1〇8並開始 分別計算該等異常得分。 接下來,判斷裝置110如步驟220與步驟230所述, 在一預定週期内,例如在兩天内判斷第一機台之一異常得 分總和是否大於一 η值,並決定是否通知使用者。如前所 述,此η值可4艮據要求調整,而在第一實施例中,η值等 於2。另外,在本發明中之預定週期係可根據製程及機台 種類、機台產出量、機台停機時間、或機台預防性保養循 環調整之要求調整。在本第一較佳實施例中,第一機台由 於發生了兩種類型的缺陷,其異常得分總和大於2,故機 台異常之自動化追蹤及通報系統100係如步驟240所述, 15 1363963 將此第一機台以及相關之該等異常資訊利用自動通報子系 統112通報予使用者,如線上製程工程師或品管工程師。 當該等異常得分總和小於2,甚至在超過2天仍小於2時, 則判定為單一事件(single event),而不列入下次預定週期内 之異常得分計算。但此一異常資訊仍儲存於機台狀態資料 庫106中作為之後異常機台之協助診斷參考資料,直到使 用者認為該等異常資訊及其異常得分可刪除。另外,在超 過預定週期之後,該等異常得分亦可依使用者規定歸零, 以避免影響後續的異常得分計算;當然該等異常資訊仍係 儲存於機台狀態資料庫106中,以作為日後異常機台之協 助診斷參考資料。 由於本發明所提供之機台異常之自動化追蹤及通報系 統100與方法係自動擷取不同產品、不同製程、不同機台 之檢測結果或註解,並進行分析與計算,因此即使上述缺 陷係發生於不同班次,或由不同的工程師所標示,仍可藉 由本系統自動追蹤並整合分析出可能異常機台之異常資 訊,而自動將其通報予使用者。因此,可避免人工交接可 能產生的遺漏或誤差’亦可彌補忽略微小異常可能導致的 問題。 以下為本發明根據上述第一流程步驟所提供之第二較 佳實施例。當經過第一機台完成多晶矽層微影製作之第二 16 1363963 批次產品經檢測發瑪有管制異常(ooc)之狀況發生,而被 ‘疋為暫停批並暫停下來。而後經由工程師確認第二抵次 產〇〇發生曝光失焦(defocus)等缺陷’但第二批次產品之缺 陷尚未達到報廢標準,故工程師判定可接續流程,並在釋 出第二批次產品的同時輸入一第一註解,例如:第二批次 產品有一晶粒發生失焦。 接下來第二批次產品係進行後續製程,例如再經由第 一機台完成金屬層微影(metal 1 photo)之製作,其經檢測後 亦發現有管制異常之狀況發生,而被暫停下來。而後經由 不同或相同的工程師確認第二批次產品發生曝光失焦 (defocus)以及金屬顆粒等缺陷,但尚未達到報廢標準,故 工程師判定可接續流程,並在釋出第二批次產品的同時輸 入一第二註解,例如:第二批次產品有一晶粒發生失焦, 且有金屬顆粒之缺陷。 本發明所提供之機台異常之自動化追蹤及通報系統 100係如步驟200所述,在不同的製程後,立即利用資料 擷取與分析子系統102立即分別自動擷取並分析第一註解 與第二註解内之該等異常資訊,例如利用一自動判斷字串 引擎針對註解中的擷取註解中的特殊字串,並分析整理成 如表格2所示之異常資訊·· 表格2 17 1363963 曰 期 批 次 缺陷 種類 發生 數量 發生範 圍% 致命缺陷 /非致命缺陷 機 厶 〇 機台異 常得分 發現層次 層次 得分 第 失 焦 1 N/a 致命 第 —— 1 Poly 1 photo 1 第 金 屬 顆 粒 N/a N/a 非致命 第 一 * 1 Metal 1 photo 1 第 失 焦 1 N/a 致命 第 —— 1 Metal 1 photo 1 接下來,則如步驟210所述,本發明所提供之機台異 常之自動化追蹤及通報系統100係利用載入裝置104將該 等異常資料載入並儲存於機台狀態資料庫106.中。並如步 驟220所述,利用異常資訊計算裝置108分別給予該等異 常資訊一異常得分,例如如表格2所標示在機台異常得分 一欄位予以一異常得分;並利用異常資訊計算裝置108根 據該等異常資訊與相關之可能異常機台分別計算個別可能 異常機台的異常得分。 接下來,判斷裝置110如步驟220與步驟230所述, 在預定週期内,例如兩天内判斷第一機台之一異常得分總Step 200: The automatic tracking and notification system 100 for the abnormality of the machine provided by the present invention utilizes the data acquisition and analysis subsystem 1 to automatically input the input data of the engineer, for example, the abnormal information in the annotation; The extraction and analysis subsystem can also directly extract the above detection results and integrate the analysis to obtain a plurality of abnormal information. Step 210: Continue with step 200. The loading device 104 is then used to automatically load and store the exceptions δίΐ in the machine state database. It is worth noting that the abnormal information can also be loaded and stored in the machine status database 1〇6 by manually loading the shock 104. In addition, when the new machine enters the process, the new machine's related difference > βίΐ can also be manually loaded into the machine status database 1〇6 through the loading device 1〇4. The abnormal information stored in the machine status database 106 includes at least the date, the product batch, the defect type, the defect degree, the related abnormal machine, and the abnormality score of the possible abnormal machine. Step 220: Continue with step 210. The abnormality information calculating means 1 8 calculates the abnormal payout amount of the individual month b different crane stations based on the abnormality information and the at least one related abnormal machine. Since the throughput 12 exposure 63963 is different from the machine idle time and the preventive maintenance (PM cycle), the abnormality information computing device 108 calculates the abnormal scores. The abnormal score within a predetermined period is calculated; and the abnormal score outside the period is negligible or not included in the calculation, but can still be stored in the machine state database 106 as an auxiliary diagnostic reference material for the subsequent abnormal machine. Step · Step 230: Continue with step 2 plus. The judging means 110 is used to judge whether the sum of the abnormal scores of one of the individual possible abnormal machines is greater than an η value. Dream 240: Continue with step 230. When the sum of the abnormal scores is greater than the value of η in the predetermined period, the judging means 11 judges that the abnormal abnormality machine and the associated abnormality information are notified to the at least one user by the automatic notification subsystem 112. And when the sum of the abnormal scores is less than the value of η; even if it is less than the value of η after exceeding the predetermined period, the abnormal score calculation is not included in the next predetermined period, but is still stored in the machine state database 106 as the latter Assist the diagnostic reference of the abnormal machine until the user thinks that the abnormal information and abnormal score can be zeroed or deleted. The following is a first preferred embodiment of the invention provided in accordance with the first flow steps described above. When the first batch of products produced by the first machine completed polycrystalline germanium lithography (p〇ly i ph〇t〇) was found to have a controlled abnormality (ooc), it was identified as a suspension and suspended. (4) Deficiencies such as defocus and dropped particles occur in the first batch of products through the Weicheng. If the defect is serious and it is scrapped 栌 13 1363963 The first batch is considered as a scrapped batch, and the first machine must be calibrated immediately. If the scrapping standard has not been reached and the engineer determines that the process can be continued, a note will be entered at the same time as the first batch is released. For example, the first batch of products has a grain out of focus and 50% of the particles fall. defect. The automatic tracking and notification system 100 for machine abnormalities provided by the present invention is as described in step 200. After the different processes are completed, the data acquisition and analysis subsystem 102 automatically captures and analyzes the engineer's annotations. The abnormal information is obtained, so the first batch of products receives the abnormal information as shown in Table 1: Table 1 The range of occurrence of the batch defect type in the flood season (%) Fatal defect / Non-fatal defect Machine machine abnormal score discovery level Level score No. 1 N/a Fatal 1 Poly 1 photo 1 First particle drop N/a 50 Fatal first 1 Poly 1 photo 1 It is worth noting that the data acquisition and analysis subsystem 102 is not for engineers The part of the annotation can be temporarily represented by n/a, and the machine that the batch has just experienced will be automatically retrieved, and the first machine 14 1363963 will be displayed in the abnormal information of Table 1. In addition, the user may pre-set the types of defects that may occur in different processes and different machines as killer defects or non-killer defects, while the data acquisition and analysis subsystem 102 is in accordance with the user's The predetermined rules automatically classify defects within the annotation and record them as fatal or non-fatal defects. Next, as described in step 210, the automatic tracking and notification system 100 for the machine abnormality provided by the present invention uses the loading device 104 to store the abnormal information contained in the table 1 in the machine state database 106. . And, as described in step 220, the abnormality information computing device 108 respectively gives the abnormal information an abnormality score, and an abnormality score is given to the abnormality score of the machine indicated in Table 1; the abnormal information computing device 1〇8 starts. These abnormal scores are calculated separately. Next, the determining means 110 determines whether the sum of the abnormal scores of the first machine is greater than a value of η within a predetermined period, for example, within two days, as determined by step 220 and step 230, and decides whether to notify the user. As described above, this η value can be adjusted as required, and in the first embodiment, the η value is equal to 2. In addition, the predetermined period in the present invention can be adjusted according to the requirements of the process and the type of the machine, the output of the machine, the downtime of the machine, or the preventive maintenance cycle of the machine. In the first preferred embodiment, the first machine has two types of defects, and the sum of the abnormal scores is greater than two, so the automatic tracking and notification system 100 of the machine abnormality is as described in step 240, 15 1363963 The first machine and related abnormal information are communicated to the user by the automatic notification subsystem 112, such as an online process engineer or a quality control engineer. When the sum of the abnormal scores is less than 2, even if it is less than 2 for more than 2 days, it is judged as a single event, and is not included in the abnormal score calculation in the next predetermined period. However, this abnormal information is still stored in the machine status database 106 as an auxiliary diagnostic reference material for the abnormal machine, until the user thinks that the abnormal information and its abnormality score can be deleted. In addition, after a predetermined period of time, the abnormal scores may also be zeroed according to the user's regulations to avoid affecting the subsequent abnormal score calculation; of course, the abnormal information is still stored in the machine state database 106 as a future Assisted diagnostic reference materials for abnormal machines. Since the automatic tracking and notification system 100 and the method for the abnormality of the machine provided by the present invention automatically extract test results or annotations of different products, different processes, different machines, and perform analysis and calculation, even if the above defects occur in Different shifts, or marked by different engineers, can still automatically track and analyze the abnormal information of the abnormal machine through the system, and automatically report it to the user. Therefore, it is possible to avoid the omission or error that may be caused by manual handover, which can also compensate for the problems that may be caused by neglecting minor anomalies. The following is a second preferred embodiment of the present invention provided in accordance with the above first flow steps. When the second 16 1363963 batch product of the polycrystalline germanium lithography was completed by the first machine, the condition of the controlled abnormality (ooc) was detected, and was paused and suspended. Then, the engineer confirmed that the second offset was defocus and other defects, but the defects of the second batch of products have not reached the scrapping standard, so the engineer judged that the process could be continued and released the second batch of products. At the same time, a first annotation is input, for example, a second batch of products has a grain defocusing. The second batch of products is followed by a subsequent process. For example, the metal 1 photo is produced through the first machine. After the test, it is found that a controlled abnormality occurs and is suspended. Then, through different or the same engineer, the defects of the second batch of products, such as defocus and metal particles, were confirmed, but the scrapping standards were not met. Therefore, the engineer determined that the process could be continued and released the second batch of products. A second annotation is entered, for example, a second batch of products has a grain that is out of focus and has defects in the metal particles. The automatic tracking and notification system 100 for the machine abnormality provided by the present invention, as described in step 200, immediately uses the data acquisition and analysis subsystem 102 to automatically capture and analyze the first annotation and the first, respectively, after different processes. The abnormal information in the two annotations, for example, using an automatic judgment string engine for the special character string in the annotation in the annotation, and analyzing and arranging into the abnormal information as shown in Table 2. Table 2 17 1363963 The number of occurrences of batch defect types is %% Fatal defect/non-fatal defect Machine 异常 Machine abnormal score Find hierarchical level score Defocus 1 N/a Fatal - 1 Poly 1 photo 1 Metal particle N/a N/ a Non-fatal first* 1 Metal 1 photo 1 Defocus 1 N/a Fatal - 1 Metal 1 photo 1 Next, as described in step 210, the automated tracking and notification of machine anomalies provided by the present invention The system 100 loads and stores the abnormal data in the machine state database 106 using the loading device 104. And, as described in step 220, the abnormality information computing device 108 respectively gives the abnormality information an abnormality score, for example, an abnormality score is given in the field of the abnormality score of the machine as indicated in Table 2; and the abnormality information computing device 108 is used according to the abnormality information The abnormality information and the related abnormal machine are respectively calculated for the abnormality scores of the individual possible abnormal machines. Next, the determining device 110 determines, according to step 220 and step 230, that the abnormal score of one of the first machines is total within a predetermined period, for example, two days.

1B 1363963 和是否大於一 η值,並決定是否通知使用者。如前所述, 此η值可根據要求調整,而在第二實施例中,η值等於2。 在本第二較佳實施例中,第一機台由於在2天内發生了兩 種類型的缺陷,且其異常得分總和等於3,故機台異常之 自動化追蹤及通報系統100係如步驟240所述,立即將此 第一機台以及相關之該等異常資訊通過自動通報子系統 112通報予使用者,如線上製程工程師或品管工程師。如 前所述,而當該等異常得分總和小於2,或在在超過2天 仍小於2時,則判定為單一事件,而不列入下次預定週期 内之異常得分計算,但仍儲存於機台狀態資料庫106中作 為之後異常機台之協助診斷參考資料,直到使用者認為該 等異常資訊及異常得分可刪除。 在第二較佳實施例中,非致命缺陷亦列入異常得分計 算之範圍内,然而使用者係可根據不同製程與不同產品之 容忍度要求,將非致命缺陷定義為不列入異常得分計算或 經過加權後才列入異常得分計算,以避免過分放大非致命 缺陷對於機台之影響而造成不必要的成本支出,同時亦可 避免過分忽視非致命缺陷的影響’使得非致命缺陷於後續 製程中引發其他缺陷而造成後續重大的異常,導致產品良 率與生產成本的損失。 值得注意的是,由於本發明所提供之機台異常之自動 19 1363963 化追蹤及通報系統100與方法係自動擷取不同製程之檢測 結果或註解,並進行分析與計算,因此即使缺陷係發生於 不同製程、不同膜層、或不同時間,而由不同的工程師所 標示,仍可藉由本系統自動追蹤出.同一異常機台之異常資 訊,而將其通報予使用者。 請同時參閱第1圖與第3圖,其中第3圖係為本發明 所提供之異常機台之自動化追蹤及通報之方法之一第二流 程示意圖。如前所述,當使用者如線上製程工程師或品管 工程師針對不同製程之不同產品進行檢測,而獲得如前所 述之原始資料後,線上製程工程師或品管工程師會根據此 原始資料分析而得到一檢測結果,並判別某一批次之產品 是否可放行。接下來,如第3圖所示,本發明所提供之機 台異常之自動化追蹤及通報之方法包含有以下步驟: 步驟300 :利用資料擷取與分析子系統1〇2自動擷取工程 師之輸入資料,例如註解内之該等異常資訊;資料擷取 與分析子系統亦可直接自動擷取上述檢測結果或線上製 程數據、離線製程數據、機台監測數據、或產品缺陷數 據,並整合分析而得到複數個異常資訊,並根據該等異 常資訊診斷出至少一可能異常機台。 步驟310 :接續步驟300。隨後載入裝置104係用以將該等 異常資訊自動載入並儲存於機台狀態資料庫106中。如 前所述,該等異常資訊亦可藉由人工經由載入裝置104 20 載入及儲存於機台狀態資料庫綱。另外,冬一新機么 時,亦可透過載人裝置1Q4人工‘該新機口台 ^」貝讯於機台狀態f料庫1〇6心錯存於機台狀態 請庫伽之異常資關至少包含有日期、產品批次、 缺陷種類、缺_度、可能異常機台、該可能異常機台 之異常得分等資料。 步 驟320 :接續步驟310。利用異常資訊計算裝置灌根據 該異常資訊與相關之可能異常機台分別計算個別異常機 台在一預定週期内之異常得分。 值得注意的是,由於各機台產出與停機時間、預防性 保養循環皆不同’因此該異常資訊計算裝置⑽在計算該 異常得分時’料算該預定·内之異常得分;而該預定 週期外之異常得分則可忽略或儲存於機台狀態資料庫⑽ 中作為之後協助異常機台之診斷參考資料。 步驟330:接續步,驟320。利用判斷裝f ιι〇來判斷個別可 能異常機台之一異常得分總和是否大於一 n值。 步驟340 ··接續步驟。當該異常得分總和在該預定週期 内大於該η值時,即利用自動通報子系統112將該可能 異常機台與相關之該等異常資訊通報予至少一使用者。 而當該異常得分總和在超過該預定週期之後仍小於〇值 時,則不列入下次預定週期内之異常得分計算,但仍儲 存於機台狀態資料庫106中作為之後異常機台之協助診 211B 1363963 and whether it is greater than a value of η, and decide whether to notify the user. As previously mentioned, this η value can be adjusted as required, while in the second embodiment, the η value is equal to two. In the second preferred embodiment, the first machine has two types of defects in 2 days, and the sum of the abnormal scores is equal to 3, so the automatic tracking and notification system 100 of the machine abnormality is as shown in step 240. As described, the first machine and related abnormal information are immediately notified to the user through the automatic notification subsystem 112, such as an online process engineer or a quality control engineer. As mentioned above, when the sum of the abnormal scores is less than 2, or is still less than 2 in more than 2 days, it is judged as a single event, and is not included in the abnormal score calculation in the next predetermined period, but is still stored in The machine status database 106 serves as an auxiliary diagnostic reference material for the subsequent abnormal machine until the user thinks that the abnormal information and the abnormality score can be deleted. In the second preferred embodiment, the non-fatal defect is also included in the calculation of the abnormal score. However, the user can define the non-fatal defect as not included in the abnormal score calculation according to the tolerance requirements of different processes and different products. Or after weighting, the abnormal score calculation is included to avoid excessively amplifying the impact of non-fatal defects on the machine, resulting in unnecessary cost, and avoiding excessive neglect of the effects of non-fatal defects. Other defects are caused in the subsequent major anomalies, resulting in loss of product yield and production costs. It is worth noting that since the automatic 19 1363963 tracking and notification system 100 and the method provided by the present invention automatically capture test results or annotations of different processes, and perform analysis and calculation, even if the defect occurs in Different processes, different layers, or different times, and marked by different engineers, the system can automatically track the abnormal information of the same abnormal machine and notify it to the user. Please refer to Fig. 1 and Fig. 3 at the same time. Figure 3 is a second flow diagram of the method for automatic tracking and notification of the abnormal machine provided by the present invention. As mentioned above, when a user, such as an online process engineer or a quality control engineer, tests different products of different processes and obtains the original data as described above, the online process engineer or quality control engineer analyzes the original data. Obtain a test result and determine whether a batch of products can be released. Next, as shown in FIG. 3, the method for automatically tracking and reporting the abnormality of the machine provided by the present invention comprises the following steps: Step 300: automatically inputting the input of the engineer by using the data acquisition and analysis subsystem 1〇2 Data, such as such anomalous information in the annotation; the data acquisition and analysis subsystem can also directly retrieve the above detection results or online process data, offline process data, machine monitoring data, or product defect data, and integrate analysis Obtaining a plurality of abnormal information, and diagnosing at least one possible abnormal machine according to the abnormal information. Step 310: Continue step 300. The load device 104 is then used to automatically load and store the exception information in the machine status database 106. As described above, the abnormal information can also be loaded and stored in the machine state database through the loading device 104 20 by hand. In addition, when the new machine is used in winter, it can also be manually used by the manned device 1Q4. The new machine mouth station ^"Bai Xun in the machine state f library 1 〇 6 heart error in the machine state, please Kuga The information includes at least the date, product batch, defect type, lack of _ degree, possible abnormal machine, abnormal score of the possible abnormal machine. Step 320: Continue with step 310. The abnormality information computing device is used to calculate the abnormal scores of the individual abnormal machines in a predetermined period according to the abnormal information and the related abnormal machine. It is worth noting that since the output of each machine is different from the downtime and the preventive maintenance cycle, the abnormal information computing device (10) calculates the abnormal score in the predetermined time when calculating the abnormal score; and the predetermined period The abnormality score outside can be ignored or stored in the machine status database (10) as a diagnostic reference for assisting the abnormal machine. Step 330: Continue step, step 320. Use the judgment f ιι〇 to determine whether the sum of the abnormal scores of one of the possible abnormal machines is greater than an n value. Step 340 ·· Follow-up steps. When the sum of the abnormal scores is greater than the η value within the predetermined period, the automatic notification subsystem 112 is used to notify the at least one user of the possible abnormal machine and the associated abnormal information. When the sum of the abnormal scores is still less than the threshold after exceeding the predetermined period, the abnormal score calculation in the next predetermined period is not included, but is still stored in the machine state database 106 as the assistance of the abnormal machine afterwards. Diagnosis 21

lDOJyOJ lDOJyOJ 該異常資訊與該異常得分 斷參考資料,直到使用者認為 可刪除。 乂下為本發明根據上 佳實施例。當一铤㈣第-流粒步驟所k供之弟三較 二批次I 機台完成多晶矽層微影製作之第 被認定為二/則發?有管制異常(ooc)之狀況發生,而 A洚口鉻τ比並暫停下來。而後經由工程師確認第三批 ㈣與微粒掉落等缺陷,但尚未達到報廢 :η 4師判定可接續流程,則會在釋出第三批次 斗二认一第一註解,例如:第三批次產品有-晶粒發 生失焦。 本毛月所提供之機台異常之自動化追縱及通報系統 100係如步驟300所述,在不同的製程後,立即利用資米 榻取與分析子系統1G2立即分別自動錄並純註解内: 該等異常資訊’或者直㈣取第三批次產品相關之線幻 程數據、離線製程數據、機台監測數據等、或產品缺顯』 據’分析整理成如表格3所示之異常資訊,並指認出至/ -相關之可能異常機台。例如在本第三較佳實施例中,: 用者雖未於轉中朗第三批次之缺陷係來自於哪一機 台’但由於機台狀態資料庫1〇6係可提供龐大的異常機爸 診斷參考資訊,因此資料_與分析子系統而係可_ 使用者所輸人、或可根據上述其他檢測結果自動分析而相 22 1363963 認診斷出第三批次產品所發生之缺陷係導因於第一機台, 並自動於表格3中之機台攔位中紀錄為第一機台。 表格3 曰 期 批 次 缺陷 麵 發生 數量 發生 範圍% 致命缺陷 /非致命缺 陷 機 台 機台異 常得分 發現層次 層次 得分 第 失焦 1 N/a 致命 第 —一 氺 1 Poly 1 photo 1 第 微粒 N/a 50 致命 第 — 氺 1 Poly 1 photo 1 接下來,則如步驟310所述,利用載入裝置104將該 等異常資料載入並儲存於機台狀態資料庫106。並如步驟 320所述,利用異常資訊計算裝置108分別給予該等異常 資訊一異常得分,例如如表格3所標示在機台異常得分一 欄位予以一異常得分。隨後異常資訊計算裝置108係根據 該異常資訊與相關之可能異常機台分別計算個別可能異常 機台之異常得分。 接下來,判斷裝置110如步驟320與步驟330所述, 23 1363963lDOJyOJ lDOJyOJ The exception information and the exception score are cut off until the user thinks it can be deleted. The present invention is based on the preferred embodiment of the invention. When the first (four) first-flowing step is provided for the third generation, the second batch of I machine is completed, and the third layer of the lithography is determined to be the second/then. A condition of a controlled abnormality (ooc) occurs, and the A-port chrome tau ratio is suspended. Then the engineer confirmed the third batch (4) and the defects such as falling particles, but it has not yet reached the end of life: η 4 division judged that the process can be continued, and the third batch of the second batch of the first comment is released, for example: the third batch The secondary product has - the grain is out of focus. The automatic tracking and notification system 100 for the abnormality of the machine provided by Maoyue is as described in step 300. Immediately after the different processes, the 1M2 of the metering and analysis subsystem is automatically recorded and automatically recorded in the following: Such abnormal information 'or straight (4) take the third batch of product-related line magic data, offline process data, machine monitoring data, etc., or product lack of display" according to the analysis of the abnormal information as shown in Table 3, And identify the possible abnormal machine to /. For example, in the third preferred embodiment, the user does not have a defect in the third batch of the transfer to which the machine is from the machine. However, since the machine state database 1〇6 system can provide a huge abnormality. The dad diagnoses the reference information, so the data_and the analysis subsystem can be _ user input, or can be automatically analyzed according to the other test results described above. 22 1363963 diagnoses the defects of the third batch of products Because of the first machine, and automatically recorded in the machine block in Table 3 as the first machine. Table 3 Range of occurrences of defective batch defects in the flood season % fatal defect / non-fatal defect machine machine abnormality score found level of hierarchy score defocus 1 N / a fatal - one 氺 1 Poly 1 photo 1 first particle N / a 50 Fatal - 氺 1 Poly 1 photo 1 Next, the abnormal data is loaded and stored in the machine state database 106 by the loading device 104 as described in step 310. And as described in step 320, the abnormality information computing device 108 respectively gives the abnormal information an abnormality score, for example, an abnormality score is given in the field of the abnormality score of the machine as indicated in Table 3. The abnormality information computing device 108 then calculates the abnormality scores of the individual possible abnormal machines based on the abnormality information and the associated possible abnormal machine. Next, the determining device 110 is as described in step 320 and step 330, 23 1363963

在—預定週_,例如兩天内判斷第—機台之—显常得分 總和是否大於等於n,並決定是否通知使用者。如前所述, ^值可根據要求調整,而在第三較佳實施例中,n值係 專於h另外在本第三㈣實施财定義為兩天之預定週 期係可根據製程及機台種類、機台產出量、機台停機時間、 ^機台預防性保養循環調整之要求調整。在本第三較佳實 施例中’第—機台由於在2天内發生了二種類型的缺陷, 其異常得分總和大於2,故機台異常之自動化追蹤及通報 糸統100係如步驟340戶斤述,將此第一機台與相關之該等 異常資訊利用自動通報子系統H2通報予使用者,如線上 製程工程師或品管工程師。如前所述,而當該等異常得分 積分總和小於2,或在超過2天仍小於2時,則判定為單 一事件,而不列入下次預定週期内之異常得分計算,但仍 儲存於機台狀態資料庫1〇6中作為之後異常機台之協助診 斷參考"k料,直到使用者認為該異常資訊與該異常得分可 刪除。 在第三較佳實施例中,本發明所提供之機台異常之自 動化追縱及通報系統100與方法係藉由儲存於機台狀態資 料庫106中之大量異常機台診斷參考資料自動分析並指認 出至少一可能之異常機台。 以下為本發明根據上述第二流程步驟所提供之第四較 24 1363963 佳實施例。當一經過第一機台完成多晶矽層微影製作之第 四批次產品經檢測發現有管制異常之狀況發生,而工程師 確認第四批次產品發生曝光失焦缺陷,但尚未達到報廢標 準,因此工程師判定可接續流程,則會在釋出第四批次產 品的同時輸入一第一註解,例如:第四批次產品有一晶粒 發生失焦。 接下來一第五批次產品亦經由第一機台完成金屬層微 影之製作。而其經檢測後亦發現有管制異常之狀況發生, 而被暫停下來。而後經由不同或相同的工程師確認第五批 次產品發生曝光失焦以及金屬顆粒等缺陷,但尚未達到報 廢標準,故工程師判定可接續流程,並會在釋出第五批次 產品的同時輸入一第二註解,例如:第五批次產品有一晶 粒發生失焦,且有金屬顆粒之缺陷。 接下來另有一第六批次貨品經由一第二機台完成多晶 矽層微影之製作。而其經檢測後亦發現有管制異常之狀況 發生,.而被暫停下來。而後經由不同或相同的工程師破認 第六批次產品發生曝光失焦缺陷’但亦未達到報廢標準’ 故工程師判定可接續流程,並會在釋出第六批次產品的同 時輸入一第三註解,例如:第六批次產品有一晶粒發生失 焦0 25 1363963 本發明所提供之機台異常之自動化追蹤及通報系統 100係如步驟300所述,在不同的製程後,立即利用資料 擷取與分析子系統102立即分別自動擷取並分析第一、第 二、與第三註解内之該等異常資訊,或者直接擷取此三批 次產品相關之線上製程數據、離線製程數據、機台監測數 據等,並分析整理成如表格4所示之異常資訊: 表格4 曰 期 批 次 缺陷 纖 發生 數量 發生範 圍% 致命缺陷 /非致命缺 陷 機 台 機台異 常得分 發現層次 層次 得分 第 四 失 焦 1 N/a 致命 第 —一 1 Poly 1 photo 1 第 五 失 焦 1 N/a 致命 第 _ — 1 Metal 1 photo 1 第 五 金 屬 微 粒 N/a N/a 非致命 第 1 Metal 1 photo 1 第 六 失 焦 1 N/a 致命 第 —_ 1 Poly 1 photo 1 26 1363963 接下來,則如步驟310所述,利用載入裝置104將該 等異常資料載入並儲存於機台狀態資料庫106中。並如步 驟320所述,利用異常資訊計算裝置108給予並計算該等 異常資訊之一異常得分,例如如表格4所標示在機台異常 得分一欄位分別予以一異常得分。 接下來判斷裝置110如步驟320與步驟330所述,在 I 預定週期内,例如2天内判斷第一機台之一異常得分總和 是否大於等於2,並決定是否通知使用者。如前所述,此 一預定週期係可根據製程及機台種類、機台產出量、機台 停機時間、或機台預防性保養循環調整之要求調整。在本 第四較佳實施例中,第一機台由於在二天内發生了兩種類 型的缺陷,其異常得分總和大於2 ’故機台異常之自動化 追蹤及通報系統100係如步驟340所述,將此第一機台與 相關之異常資訊利用自動通報子系統112通報予使用者’ _ 如線上製程工程師或品管工程師。如前所述,當該等異常 得分總和小於2,或在超過2天仍小於2時,則判定為單 一事件,而不列入下次預定週期内之異常得分計算,但仍 儲存於機台狀態資料庫106中作為之後異常機台之協助診 斷參考資料,直到使用者認為該異常資訊與異常得分可刪 除。 在第四較佳實施例中,第四批次與第五批次所經歷之 27 第-機台因異常得分總和大於2,故此異常機台與相關之 異常2係由自動通報子系統112通報予至少一使用者。 广、是兒本發明所提供之自動化追縱及通報系統刚係 動追蹤㈤彳成異常機台之異常資訊,因此即使等異常 一貝訊係發生於不同製程.、不_層甚或不同批次,且由 Z的工程師所標示,仍可藉由本系統所追蹤發現,而將 其通報予使用者。 ^卜值得注意較,在h難實施财,第四批次 二/、批-人的缺陷係分別為使用者診斷為第—機台與第二 常貝訊並經由資料鞠取與分析子系統102顯示 二鉻4 ”。然而由於此二批次貨品同在第-層多晶矽層 白二了失焦的缺陷因此本發明所提供之自動化追蹤及 通糸、先100係可利用資料榻取與分析子系統搬自動分 析該等異常資訊並崎兩批次貨品之製造流程,若發現此 歷—第2機台’則資料擷取與分析子系統 係可額外指認出此第三相關異常機台,並通知使用者 :額外確5忍第三相關機台之資訊。若根據異㈣訊額外指 ^出之第二相關機台之異常得分總和大m清裝置ιι〇 p判疋須立即通知使用者,且利用自動通報子系統n2通 知使用者。 在第四fe佳實施例中,本發明所提供之機台異常之自 28 1363963 動化追蹤及通報系統100與方法係跨越不同批次、不同製 程、不同班次工程師,而自動追蹤並追認出異常資訊。更 值得注意的是,藉由儲存於機台狀態資料庫106中之大量 異常機台診斷參考資料自動分析比對,而可針對不同機台 發生的相同缺陷類型往前推論並額外追認出至少一可能之 相關異常機台,提供使用者參考。 以下為本發明根據上述第二流程步驟所提供之第五較 佳實施例。當一完成多晶矽層微影製作之第七批次產品經 檢測發現有管制異常之狀況發生,例如發生曝光失焦缺陷 等,本發明所提供之機台異常之自動化追蹤及通報系統1〇〇 係如步驟300所述,在完成多晶矽層微影製作後,立即利 用資料擷取與分析子系統102分別自動擷取並分析工程師 之註解或擷取此相關之原始資料,並分析整理成如表格5 所示之異常資訊,並指認出相關之可能異常機台。例如在 本第五較佳實施例中,曝光失焦之缺陷係對應於當站的第 一機台,例如一微影機台的曝光異常;或肇因於第二機台’ 如一多晶矽層沉積機台的微粒異常。因此資料擷取與分析 子系統102係可根據使用者所輸入、或可根據上述其他檢 測結果自動分析而指認診斷出第七批次產品所發生之缺陷 係相關於第一機台與第二機台,並自動於表格5中之機台 欄位中紀錄為該第一與第二機台。 29 1363963 , 另有一第八批次、第九批次、第十批次與第十一批次 產品分別在不同的微影機台完成多晶矽層微影製作後,皆 發現有曝光失焦缺陷。該曝光失焦缺陷係相關於不同的當 站微影機台,如一第三機台、第四機台、第五機台、與第 六機台。然本發明所提供之機台異常之自動化追蹤及通報 系統100係自動比對前該等批次產品之製作流程,而分析 出該等批次產品皆經過前述第二機台,因此資料擷取與分 Φ 析子系統102係自動於表格5中之機台欄位中分別紀錄該 等機台:In the predetermined week _, for example, two days, it is judged whether or not the total score of the first machine is equal to or greater than n, and whether or not the user is notified. As described above, the value of ^ can be adjusted according to requirements. In the third preferred embodiment, the value of n is specific to h. In addition, the third (four) implementation of the financial definition is defined as a two-day predetermined period, which can be based on the process and the machine. Type, machine output, machine downtime, and adjustment of the preventive maintenance cycle adjustment of the machine. In the third preferred embodiment, the first machine has two types of defects within two days, and the sum of the abnormal scores is greater than two, so the automatic tracking and notification system of the machine abnormality is as follows: It is said that the first machine and related abnormal information are notified to the user by the automatic notification subsystem H2, such as an online process engineer or a quality control engineer. As mentioned above, when the sum of the scores of the abnormal scores is less than 2, or is still less than 2 for more than 2 days, it is determined as a single event, and is not included in the abnormal score calculation in the next predetermined period, but is still stored in The machine status database 1〇6 is used as a reference for assisting diagnosis of the abnormal machine afterwards, until the user thinks that the abnormal information and the abnormal score can be deleted. In a third preferred embodiment, the automated tracking and notification system 100 and method for machine anomaly provided by the present invention are automatically analyzed by a large number of abnormal machine diagnostic reference data stored in the machine state database 106. Identify at least one possible abnormal machine. The following is a preferred embodiment of the fourth aspect of the present invention provided by the second flow step of the present invention. When a fourth batch of products that have been completed by the first machine to complete the polysilicon layer lithography has been found to have a controlled abnormality, and the engineer confirms that the fourth batch of products has an exposure defocus defect, but has not yet reached the scrapping standard, If the engineer determines that the process can be continued, a first note will be entered at the same time as the fourth batch of product is released. For example, a grain of the fourth batch has a defocus. The next fifth batch of products is also finished with the metal layer lithography through the first machine. After the test, it was found that a condition of controlled abnormality occurred and was suspended. Then, through different or the same engineer, the fifth batch of products was confirmed to have defects such as exposure defocus and metal particles, but the scrapping standard has not yet been reached. Therefore, the engineer judges that the process can be continued, and will input one at the same time as the fifth batch of products is released. Second note, for example, the fifth batch of products has a grain defocusing and defects in metal particles. Next, a sixth batch of goods was produced through a second machine to complete the production of polycrystalline germanium lithography. However, after the test, it was found that a condition of controlled abnormality occurred and was suspended. Then, through the different or the same engineer, the sixth batch of products was diagnosed as having an exposure defocus defect 'but the scrapping standard was not met'. Therefore, the engineer determined that the process could be continued, and the third batch of products was released while entering a third. Note: For example, the sixth batch product has a grain defocusing 0 25 1363963 The automatic tracking and notification system 100 for the machine abnormality provided by the present invention is as described in step 300, and the data is used immediately after different processes. The acquisition and analysis subsystem 102 automatically captures and analyzes the abnormal information in the first, second, and third annotations, or directly captures the online process data, offline process data, and machine related to the three batches of products. Monitoring data, etc., and analyzing and arranging the abnormal information as shown in Table 4: Table 4 The number of occurrences of batch defects in the flood season is %. The fatal defect/non-fatal defect machine machine abnormal score is found to be the fourth level of the hierarchical level score. Coke 1 N/a Fatal - 1 Poly 1 photo 1 Fifth Defocus 1 N/a Fatal _ — 1 Metal 1 photo 1 Hardware Particle N/a N/a Non-fatal 1st Metal 1 photo 1 6th out of focus 1 N/a Deadly -_ 1 Poly 1 photo 1 26 1363963 Next, as described in step 310, using the loading device 104 The exception data is loaded and stored in the machine status database 106. And as described in step 320, the abnormality information computing device 108 is used to give and calculate an abnormality score of one of the abnormal information, for example, an abnormality score is respectively given in the field of the abnormality score of the machine as indicated in Table 4. Next, the judging device 110 determines whether the sum of the abnormality scores of one of the first machines is greater than or equal to 2 within a predetermined period of I, for example, as described in step 320 and step 330, and determines whether to notify the user. As mentioned earlier, this predetermined period can be adjusted according to the requirements of the process and machine type, machine output, machine downtime, or machine preventive maintenance cycle adjustment. In the fourth preferred embodiment, the first machine has two types of defects in two days, and the sum of the abnormal scores is greater than 2'. The automatic tracking and notification system 100 of the machine abnormality is as described in step 340. The first machine and related abnormal information are notified to the user by the automatic notification subsystem 112, such as an online process engineer or a quality control engineer. As mentioned above, when the sum of the abnormal scores is less than 2, or is still less than 2 in more than 2 days, it is judged as a single event, and is not included in the abnormal score calculation in the next predetermined period, but is still stored in the machine. The status database 106 serves as an auxiliary diagnostic reference for the subsequent abnormal machine until the user believes that the abnormal information and the abnormal score can be deleted. In the fourth preferred embodiment, the sum of the abnormality scores of the 27th-machines experienced by the fourth batch and the fifth batch is greater than 2, so the abnormal machine and the associated abnormality 2 are notified by the automatic notification subsystem 112. To at least one user. Guang, is the automatic tracking and notification system provided by the invention. It is just tracking the abnormality of the abnormal machine. Therefore, even if the abnormal one is in different processes, no layer or even different batches. And marked by the engineer of Z, can still be notified to the user through the tracking of the system. ^ Bu is worth noting, in the difficult implementation of wealth, the fourth batch of two /, batch-human defect system for the user diagnosed as the first machine and the second often Beixun and through the data acquisition and analysis subsystem 102 shows dichrome 4". However, since the two batches of goods are in the same state as the first layer of polycrystalline germanium layer, the auto-tracking and overnight, first 100 series available data couching and analysis are provided. The subsystem automatically analyzes the abnormality information and the manufacturing process of the two batches of goods. If the calendar-the second machine is found, the data acquisition and analysis subsystem can additionally identify the third related abnormal machine. And inform the user: the additional 5 tolerate the information of the third related machine. If the abnormality score of the second related machine is additionally indicated according to the different (four) news, the user must notify the user immediately. And the user is notified by the automatic notification subsystem n2. In the fourth embodiment, the machine provided by the present invention is abnormal from 28 1363963. The dynamic tracking and notification system 100 and the method system span different batches and different processes. Different shift engineers The tracking and tracking of the abnormal information is automatically tracked. More importantly, the same defect type that can be generated for different machines can be automatically analyzed by a large number of abnormal machine diagnostic reference data stored in the machine state database 106. Inferring and additionally recognizing at least one possible related abnormal machine, providing a user reference. The following is a fifth preferred embodiment of the present invention according to the second flow step described above. When the polysilicon layer lithography is completed The seventh batch of products is found to have a controlled abnormal condition, such as an occurrence of exposure defocusing defects, etc., and the automatic tracking and notification system for the abnormality of the machine provided by the present invention is completed as described in step 300, and the polycrystalline silicon is completed. After the layer lithography is produced, the data acquisition and analysis subsystem 102 is automatically used to automatically extract and analyze the engineer's annotations or retrieve the relevant original data, and analyze and organize the abnormal information as shown in Table 5, and identify the abnormal information. Related to the abnormal machine. For example, in the fifth preferred embodiment, the defect of the exposure defocus corresponds to the first machine of the station. For example, an exposure abnormality of a lithography machine; or due to a particle abnormality of the second machine 'such as a polysilicon layer deposition machine. Therefore, the data acquisition and analysis subsystem 102 can be input according to the user, or can be based on The other test results are automatically analyzed and the defects diagnosed in the seventh batch of products are related to the first machine and the second machine, and are automatically recorded in the machine field in Table 5 as the first The second machine. 29 1363963, another eighth batch, the ninth batch, the tenth batch and the eleventh batch of products were respectively produced in different lithography machines after the completion of polycrystalline germanium lithography. Exposure defocusing defect. The exposure defocusing defect is related to different station lithography machines, such as a third machine, a fourth machine, a fifth machine, and a sixth machine. However, the present invention provides The automatic tracking and notification system of the machine abnormality 100 automatically compares the production processes of the previous batches of products, and analyzes that the batch products pass through the second machine, so the data acquisition and sub-measurement subsystem 102 series automatically in the machine in Table 5 Field, respectively, and other records of the machine:

表格5 曰 期 批 次 缺陷 麵 發生 數量 發生範 圍% 致命缺陷 /非致命缺 陷 機 台 機台異 常得分 發現層次 層次 得分 第 失 1 N/a 致命 第 1 Poly 1 1 七 焦 — photo 第 失 1 N/a 致命 第 1 Poly 1 1 七 焦 二 photo 第 失 1 N/a 致命 第 1 Poly 1 1 八 焦 二 photo 第 失 1 N/a 致命 第 1 Poly 1 1 八 焦 三 photo 第 失 1 N/a 致命- 第 1 Poly 1 1 photo 30 1363963 九 焦 ^·— S----— _ 第 失 1 N/a 致命 第 1 ' P〇ly 1 1 九 焦 四 Photo 第 失 1 N/a 致命 第 ---- 1 十 焦 二 Photo 第 失 1 N/a 致命 第 f〇bnr~ — 1 十 焦 五 Photo 弟 失 1 N/a 致命 第 1 ' Poly 1 1 十 焦 '—' Photo 第 失 1 N/a 致命 第 1 —------ 1 十 — 焦 六 Photo ~~~~—-. —-------Table 5 Number of occurrences of batch defects on the surface of the flood season % fatal defect / non-fatal defect machine machine abnormal score found level of hierarchy score missing 1 N / a fatal 1 Poly 1 1 7 jog - photo 1st loss N / a Fatal 1st Poly 1 1 Seven Jiao 2 photo Lost 1 N/a Deadly 1 Poly 1 1 Eight-focus II Photo Lost 1 N/a Deadly 1 Poly 1 1 Eight-focus three photo Lost 1 N/a Deadly - 1st Poly 1 1 photo 30 1363963 九焦^·— S----- _ 1 lost 1 N/a fatal 1 ' P〇ly 1 1 9 焦四Photo Lost 1 N/a Fatal -- -- 1 十焦二Photo Lost 1 N/a Fatal f〇bnr~ — 1 Ten Jiao Five Photo Brother Lost 1 N/a Fatal 1 ' Poly 1 1 Ten Jiao'—' Photo Lost 1 N/a Fatal 1st ------- 1 10 - Jiaoliu Photo ~~~~--.

該等異常資料係如步驟31〇所述,係利用載入裝置ι〇4 將載入並儲存於機台狀態資料庫應中。並如步驟細所 述,利用異常資訊計算裝置108根據該等異常資訊與相關 異常機台分縣予並計算個別異常機台之異常得分,例如 如表格5所標示在機台異常得分—欄位分料以—異常得 分0 與步驟330所述,在 之一異常得分總和是 接下來判斷裝置110如步驟320 預定週期内,例如2天内判斷各機台 川:>处3 . 否大於等於一 n值,並決定是否通知使用I。如前所述, it匕 預定週期係可根據製程及機台種類、機台產出量、機 口如機時間、或機台預防性保養循環調整之要求調整。值 侍皮意的是,在第五較佳實施例中,第>機台之異常資訊 系由本系統追蹤而指認出來的,為避免過分放大經由系統 追蹤出來的機台異常資訊而造成不必要的成本支出’在本 第五較佳實施例中,n值係設定為等於4。由表袼5可知, • 第二機台之異常得分積分總和係等於4,故機台異常之自 動化追縱及通報系統100係如步驟340所述’將此第二機 台與相關之異常資訊利用自動通報子系統112通報予使用 者,如線上製程工程師或品管工程師。而其他機台的異常 得分總和皆等於1,因此可判定為單一事件。 由於每種缺陷可能由不同類型機台的異常所引起’即 每種缺陷係對應於複數機台,因此各種機台異常所對應之 缺陷係可規定並建置在機台狀態資料庫1〇6 +。當缺陷發 生時,不僅當站機台可被指認出來,其相關的可能異常機 台都可被本發明所提供之機台異常之自動化追蹤及通報系 統100追認出來。且如本第五較佳實施例所述,即使經過 不同的同類麥機台完成的不同批次產品亦可藉由本發明所 提供之機台異常之自動化追蹤及通報方法比對所經過的流 程而自動追蹤出同一可能異常機台之異常資訊。因此即使 該等異常資訊係發生於非當站機台或不同批次,且由不同 32 的工程師所標示, 報予使用者。 仍可藉由本系統所追蹤發現 ,而將其通The abnormal data is loaded and stored in the machine status database by the loading device ι〇4 as described in step 31〇. And, as described in detail, the abnormality information computing device 108 calculates and calculates an abnormality score of the individual abnormal machine according to the abnormal information and the related abnormal machine, for example, the abnormality score of the machine indicated in Table 5. The distribution is determined by the abnormality score 0 and the step 330, and the sum of the abnormal scores is determined by the judging device 110 in the predetermined period of step 320, for example, within 2 days, each machine is determined to be: > at the position 3. No greater than or equal to one The value of n, and decide whether to notify the use of I. As mentioned earlier, the it匕 predetermined cycle can be adjusted according to the process and machine type, machine output, machine time, or machine preventive maintenance cycle adjustment requirements. The value of the value is that, in the fifth preferred embodiment, the abnormal information of the machine is identified by the system, and is unnecessary to avoid excessive amplification of the abnormal information of the machine tracked by the system. Cost of expenditure 'In the fifth preferred embodiment, the value of n is set equal to four. It can be seen from Table 5 that: • The sum of the abnormal scores of the second machine is equal to 4, so the automatic tracking and notification system 100 of the machine abnormality is as described in step 340, 'This second machine and related abnormal information The automatic notification subsystem 112 is used to notify the user, such as an online process engineer or a quality control engineer. The sum of the abnormal scores of other machines is equal to 1, so it can be judged as a single event. Since each defect may be caused by an abnormality of different types of machines, that is, each defect corresponds to a plurality of machines, and the defects corresponding to various machine abnormalities can be specified and built in the machine state database. +. When a defect occurs, not only when the station is identifiable, but also the associated abnormal machine can be identified by the automatic tracking and notification system 100 of the machine abnormality provided by the present invention. And as described in the fifth preferred embodiment, even if different batches of products completed by different similar types of wheat machines can be compared with the flow through the automatic tracking and notification method of the machine abnormality provided by the present invention. Automatically track out the abnormal information of the same possible abnormal machine. Therefore, even if such abnormal information occurs on a non-station machine or a different batch, and is marked by a different 32 engineer, it is reported to the user. Still can be found by the system

台。另外,當相同類型的缺陷在不同 根據本發明所提供 統與方法,異當 资却及 批次與不同機台發生時,本發明所提供之機台異常之自動 化追蹤及通報系統與方法係更可藉由分析不同批次所經歷 的製作流程’更往前追認出至少-相Μ台。也就是說Γ 本發明所提供之方法係可自動統計以及分析異常產品與異 常機台關聯性,而歸納出異常資訊以及可能異常之機台, 並即時報使用者。由於上述步驟皆係利用本發明所提供機 〇異吊之自動化追縱及通報系統所完成,因此可省去人工 操作之影響與彌補使用者經驗之不足,亦可作為新進人員 之教育訓練用途,故本發明更為一可跨越不同班次與不同 機台而提供完整的異常資訊相關性之系統與方法。 以上所述僅為本發明之較佳實施例’凡依本發明申請 33 1363963 專利範圍所做之均等變化與修飾,皆應屬本發明之涵蓋範 圍。 【圖式簡單說明】 第1圖係為本發明所提供之機台異常之自動化追蹤及通報 系統之一較佳實施例示意圖。 第2圖係為本發明所提供之機台異常之自動化追蹤及通報 之方法之一第一流程示意圖。 第3圖係為本發明所提供之機台異常之自動化追蹤及通報 之方法之一第二流程示意圖。 [ 主要元件符號說明】 100 機台異常之自動化追蹤及通報系統 102 資料擷取及分析子系統 104 載入裝置 106 機台狀態資料庫 108 異常資訊計算裝置 110 判斷裝置 112 自動通報子系統 200 利用一資料擷取與分析子系統自動擷取該註 解中之該等異常資訊 210 利用一載入裝置將該等異常資訊自動載入一 機台狀態資料庫 34 上咖963 22〇 230 240station. In addition, when the same type of defect is different in the system and method provided according to the present invention, the difference between the batch and the different machine occurs, the automatic tracking and notification system and method for the machine abnormality provided by the present invention is more By analyzing the production processes experienced by different batches, we can further identify at least the opposite stage. That is to say, the method provided by the invention can automatically count and analyze the correlation between the abnormal product and the abnormal machine, and summarize the abnormal information and the machine that may be abnormal, and immediately report the user. Since the above steps are all completed by the automatic tracking and notification system provided by the present invention, the influence of the manual operation and the deficiencies of the user experience can be saved, and the training effect of the new personnel can also be used. Therefore, the present invention is a system and method for providing complete anomaly information correlation across different shifts and different machines. The above are only the preferred embodiments of the present invention, and the equivalent variations and modifications made by the scope of the present invention are to be included in the scope of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS Fig. 1 is a schematic view showing a preferred embodiment of an automatic tracking and notification system for machine abnormalities provided by the present invention. Figure 2 is a first flow diagram showing one of the methods for automatic tracking and notification of machine abnormalities provided by the present invention. Figure 3 is a second flow diagram of one of the methods for automatic tracking and notification of machine abnormalities provided by the present invention. [Main component symbol description] 100 machine automatic tracking and notification system 102 data acquisition and analysis subsystem 104 loading device 106 machine state database 108 abnormal information computing device 110 determining device 112 automatic notification subsystem 200 utilizing The data capture and analysis subsystem automatically retrieves the abnormal information in the annotation 210. The abnormal information is automatically loaded into a machine state database 34 by a loading device. 963 22〇 230 240

300 3l〇300 3l〇

320 330 利用-異常資訊計算裝置根據該等異常資訊 與至少一相關之可能異常機台來分別計算— 預定週期内個別可能異常機台的異常得^ 利用-判斷裝置判斷個別可能異常機台之一 異常得分總和是否大於一 η值 當該異常得分總和大於該η值時,即利用—自 動通報子系統將該異常機台與相關之該等異 吊資訊通報予至少一使用者 利用一資料擷取與分析子系統自動榻取該註 解或相關線上製程數據、離線製程數據、機台 監測數據並整合成複數個異常資訊,同時診斷 出至少一相關之可能異常機台 利用一載入裝置將該等異常資訊自動载入一 機台狀態資料庫。 利用一異常資訊計算裝置根據該等異常資訊 與至少一相關之可能異常機台來分別計算一 預定週期内個另,J可能異常機台的異常得分。 利用判斷裝置判斷個別可能異常機台之一異 常得分總和是否大於一 η值。 當該異常得分總和大於該η值時,即利用一自 動通報子系統將該異常機台與相關之該等異 吊負汛通報予至少一使用者。 35 340320 330 The use-abnormal information computing device separately calculates, according to the abnormal information and at least one related abnormal machine, the abnormality of the individual abnormal machine in the predetermined period, and the determining device determines one of the individual abnormal machines. Whether the sum of the abnormal scores is greater than an η value, when the sum of the abnormal scores is greater than the η value, that is, the automatic notification subsystem notifies the abnormal machine and the related information to at least one user using a data capture And the analysis subsystem automatically takes the annotation or related online process data, offline process data, machine monitoring data and integrates into a plurality of abnormal information, and diagnoses at least one related abnormal machine using a loading device The exception information is automatically loaded into a machine status database. An abnormality information computing device is used to calculate an abnormality score of a possible abnormal machine in a predetermined period based on the abnormal information and the at least one related abnormal machine. The judging means judges whether the sum of the abnormal scores of one of the individual abnormal machines is greater than a value of η. When the sum of the abnormal scores is greater than the value of η, an automatic notification subsystem is used to notify the at least one user of the abnormal machine and the associated one. 35 340

Claims (1)

1363963 -----^ 申請專利範圍: ι· -種機台異常自動化追縱及通報系統,包含有. -資料獅與分析子线,用以自動 果並整合分析成複數個異常資訊; 數個檢測結 ,一機台狀態資料庫,用以儲存該等異常資訊. 、-異常資訊計算裝置,用以根據讀等異^資訊 — 相關之可能異常機台來分別計算個別該可 機夕― 常得分; 的異 一判斷裝置’用以判斷個別該可能異常機△之— 分總和是否大於一 n值,當該異常得分 異吊付 b m 刀',恩和大於該11值時, 即用以判斷須通知至少一使用者,其φ ^ 、τ項η值係可根攄播 台種類、機台產出量、機台停機時間、或機台預防性保養 循環(preventive maintenance,PM cycie)調整.以及 一自動通報子系統,用以將該等可能異常機台與該等異 常資訊通報予該使用者。 2.如申請專利範圍第1項所述之系統,其中檢測結果係來 自於線上(on-line)製程數據、離線(off-line)製程數據、機台 監測(machine monitor)數據、產品缺陷(product defect)數 據、或該使用者之輸入資料。 3.如申請專利範圍第1項所述之系統,其中該資料擷取與 分析子系統係用以根據該等異常資訊分析及診斷出該等可 36 063963 能異常機台。 10丨年2月13日修正替換頁 4♦如申請專利範圍第i項所述之系統,更包含—載入裝 置用以將該等異常資訊自動載入該機台狀態資料庫。 如申。月專利範圍第4項所述之系統,其中.該載入裝置係 =以人工載入該等異常資訊,或人工載入任一非來自於該· 貝料掏取及分析子m新異f f訊於該機台狀態資料 庫0 6.如申請專利範圍第丨項所述之系統,其中儲存於該機台 狀⑪料庫之該等異常資訊至少包含有日帛、產品批次、 缺種類、缺陷程度、該相關可能異常機台 '與該可能異 常機台之該異常得分料。 ” 7等二2專利範圍第1項所述之系統,其中^值係大於 8田如申請專利範圍第1項所述之系統’其中該判斷裝置係 用以簡-預定週㈣個別可能異f機台之該異常得分她 ㈣7預定週期内之該異常得分總和大於該n值時,即 4J斷須通知該使用者。 9.如申請專利範圍第8項所述之系統,其中該預定週期係 37 1363963 _:_ 10丨年2月13曰修正替換頁 可根據機台種類、機台產出量、機台停機時間、或機台預 防性保養循環調整。 10.如申請專利範圍第1項所述之系統,其中該自動通報 子系統包含有電子信件、電腦整合製造(computer integrated • manufacture,CIM)系統之警示裝置、手機簡訊、機台操作 -· 介面警示訊號、或終端使用者電腦螢幕等。. • 11. 一種機台異常自動化追蹤及通報之方法,包含有: 利用一資料擷取與分析子系統自動擷取複數個檢測結 果並整合成複數個異常資訊; 利用一機台狀態資料庫儲存該等異常資訊; 利用一異常資訊計算裝置根據該等異常資訊與至少一 相關之可能異常機台來分別計算個別可能異常機台的異常 得分; • 利用一判斷裝置判斷個別可能異常機台之一異常得分 總和是否大於一 η值,其中該η值係可根據機台種類、機 台產出量、機台停機時間、或機台預防性保養循環調整; 以及 當#ί別可能異常機台之該異常得分總和大於.η時,即判 斷利用一自動通報子系統將該可能異常機台與該等異常資 * 訊通報予至少一使用者。 38 12.如申請專利範圍第u L- 果係來自於、 其巾該等檢測結 自於線上衣域據、離線製程機台 產品缺陷數據、或該使用者之輪入資料。 據 專利範圍第n項所述之方法,其中該資料擷取 i;機台。、統係根據該等異常資訊分析及判斷出該等可能. 如申。月專利範圍第u項所述之方法,更包含一利用一鲁 載裝置將該等異常資訊自動載入該機台狀態資料庫之步 a如申請專利範圍第14項所述之方法,更包含一 利用欠該載入裳置人工載入該等異常資訊’或任—非來自於 該貝料擷取及分析子系統之新異f f訊該機台狀態資料 庫。 16·如申5月專利範圍第u項所述之方法’其中儲存於該 台狀態資料庫之該等異常資訊至少包含有日期、產品批 -人、缺陷種類、缺陷程度、該相關可能異常機台' 與該可 月b異系機台之該異常得分等資料。 17.如申請專利範圍第11項所述之方法,其中該η值係大’ 39 1363963 於等於(g) 2 101年2月13曰修正替換頁 18 .如申請專利範圍第u項所述之方 判斷裝置分析及判斷一預 更匕& 一利用該 異常得分總和,當該預定期=内個別可能異常機台之該 ?值時,即判斷須通知該使用者。之該異常得分總和大於該 19.如申請專利範圍第18 七 係可根據機台種類、機A、、方法’其中該預定週期 預防性保養循環調整。^ 1、機台停機時間、或機台 肌如申請專利範圍第u項所述之方法,其中該 二系=期電子信件、電腦整合製造系統(C Μ)之二 :機間訊、機台操作介面警示訊號 電 «幕將該異«訊通報予該使用者。電 十 圖式:1363963 -----^ Patent application scope: ι· - Anomaly automatic tracking and notification system for seed machines, including - Data lion and analysis sub-line for automatic fruit analysis and integrated analysis into multiple abnormal information; a detection node, a machine state database for storing the abnormal information. - An abnormal information calculation device for calculating the individual machine-days according to the information of the abnormal information that is related to the reading - Frequently scored; the different judgment device 'is used to determine whether the sum of the individual abnormal machine △ is greater than an n value, and when the abnormal score is different from the bm knife', when the sum is greater than the 11 value, Judging that at least one user must be notified, the η ^ and τ η values can be adjusted according to the type of broadcast station, the machine output, the machine downtime, or the preventive maintenance cycle (PM cycie). And an automatic notification subsystem for notifying the potential abnormal machine and the abnormal information to the user. 2. The system of claim 1, wherein the test results are from on-line process data, off-line process data, machine monitor data, product defects ( Product defect) Data, or input from the user. 3. The system of claim 1, wherein the data acquisition and analysis subsystem is configured to analyze and diagnose the abnormal machine according to the abnormal information. Revised replacement page on February 13, 2010. 4♦ The system described in claim i, in addition to the loading device, is used to automatically load the abnormal information into the machine status database. Such as Shen. The system of claim 4, wherein the loading device is to manually load the abnormal information, or manually load any non-from the According to the system of the state of the machine, the system described in claim 2, wherein the abnormal information stored in the machine-like 11 library includes at least a sundial, a product batch, and a missing type. The degree of defect, the relevant abnormal machine 'and the abnormality of the possible abnormal machine. The system described in Item 1 of the second and second patents, wherein the value system is greater than 8 fields, such as the system described in claim 1 of the patent application, wherein the judging device is used for the simple-predetermined week (4), and the individual may be different. The abnormality score of the machine (4) 7 when the sum of the abnormal scores in the predetermined period is greater than the value of n, that is, the 4J is required to notify the user. 9. The system of claim 8, wherein the predetermined period is 37 1363963 _:_ February 13th, 13th, the revised replacement page can be adjusted according to the type of machine, the output of the machine, the downtime of the machine, or the preventive maintenance cycle of the machine. 10. If the patent application scope is the first item The system, wherein the automatic notification subsystem comprises an electronic mail, a computer integrated manufacturing (CIM) system warning device, a mobile phone newsletter, a machine operation--interface warning signal, or a terminal user computer screen Etc.. 11. A method for automated tracking and notification of machine anomalies, including: Using a data acquisition and analysis subsystem to automatically capture multiple test results and integrate them into multiple exceptions Information; using a machine state database to store the abnormal information; using an abnormal information computing device to calculate anomaly scores of individual possible abnormal machines according to the abnormal information and at least one related abnormal machine; The judging device judges whether the sum of the abnormal scores of one of the possibly abnormal machines is greater than an η value, wherein the η value can be adjusted according to the type of the machine, the output of the machine, the downtime of the machine, or the preventive maintenance cycle of the machine; And when the sum of the abnormal scores of the abnormal machine is greater than .n, it is judged that the automatic abnormality subsystem is used to notify the at least one user of the abnormal abnormality with the automatic notification subsystem. For example, the scope of the patent application is derived from the fact that the test results from the online clothing domain data, the offline process machine product defect data, or the user's wheeled data. The method described, wherein the data is taken from the machine; and the system analyzes and judges the possibility according to the abnormal information. The method further includes a step of automatically loading the abnormal information into the machine state database by using a device, such as the method described in claim 14 of the patent scope, and further including using the The manual loading of such anomalous information 'or any - not from the bedding material acquisition and analysis subsystem, the new state information library. 16 · As stated in the May patent scope The method of storing the abnormal information stored in the state database includes at least a date, a product batch-person, a defect type, a defect degree, the related abnormal machine platform, and the abnormality of the monthly machine 17. The method of claim 11, etc. 17. The method of claim 11, wherein the η value is large '39 1363963 is equal to (g) 2 February 13, 101 revised replacement page 18. If the patent application scope is u The party judging device analyzes and judges a pre-replacement & 1 using the sum of the abnormal scores, and when the predetermined period = the value of the individual possible abnormal machine, it is judged that the user is notified. The sum of the abnormal scores is greater than the 19. The scope of the patent application can be adjusted according to the type of the machine, the machine A, and the method 'the predetermined period of the preventive maintenance cycle. ^ 1, the machine downtime, or the machine muscle as described in the patent scope of the method described in item u, wherein the second system = period e-mail, computer integrated manufacturing system (C Μ) two: machine, machine The operation interface warning signal is notified to the user. Electricity ten pattern:
TW96150638A 2007-12-27 2007-12-27 Automated abnormal machine tracking and notifying system and method TWI363963B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
TW96150638A TWI363963B (en) 2007-12-27 2007-12-27 Automated abnormal machine tracking and notifying system and method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
TW96150638A TWI363963B (en) 2007-12-27 2007-12-27 Automated abnormal machine tracking and notifying system and method

Publications (2)

Publication Number Publication Date
TW200928726A TW200928726A (en) 2009-07-01
TWI363963B true TWI363963B (en) 2012-05-11

Family

ID=44864189

Family Applications (1)

Application Number Title Priority Date Filing Date
TW96150638A TWI363963B (en) 2007-12-27 2007-12-27 Automated abnormal machine tracking and notifying system and method

Country Status (1)

Country Link
TW (1) TWI363963B (en)

Families Citing this family (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
TWI456365B (en) * 2011-04-19 2014-10-11 矽品精密工業股份有限公司 Method for managing quality abnormality, computer-readable recording medium having built-in program for quality abnormality management and computer program products
CN114254847A (en) * 2020-09-23 2022-03-29 长鑫存储技术有限公司 Data integration analysis system and data integration analysis method
US11869100B2 (en) 2020-09-23 2024-01-09 Changxin Memory Technologies, Inc. Data consolidation analysis system and data consolidation analysis method

Also Published As

Publication number Publication date
TW200928726A (en) 2009-07-01

Similar Documents

Publication Publication Date Title
CN103187343B (en) Intelligent defect diagnostic method
US6295478B1 (en) Manufacturing process change control apparatus and manufacturing process change control method
US7937168B2 (en) Automated abnormal machine tracking and notifying system and method
TW518645B (en) Method and system of automatic wafer manufacture quality control
JPH1145919A (en) Manufacture of semiconductor substrate
JP2002149860A (en) Maintenance and management method for facility in manufacturing business and maintenance and management support system
US7117057B1 (en) Yield patrolling system
CN104867840A (en) Yield enhancement (YE) on-line detection management and control method
TWI363963B (en) Automated abnormal machine tracking and notifying system and method
JP2010231338A (en) Apparatus and method for analyzing factor
CN101852741A (en) Method for defect diagnosis and management
TWI314340B (en)
JPH05253797A (en) Judging system for abnormality on line
JPWO2019016892A1 (en) Quality analyzer and quality analysis method
CN110763979A (en) LED wafer point measurement automatic system based on MES system
JP2002110493A (en) Method and apparatus for extracting failure in machining process
CN111106028A (en) Real-time monitoring method for semiconductor chip testing process
JP5138396B2 (en) Production process abnormality detection method and production process abnormality detection system, program for causing a computer to execute the production process abnormality detection method, and computer-readable recording medium storing the program
JP2008305173A (en) Maintenance device, maintenance method and program for paper sheet processor
CN108062718B (en) Processing method and processing system for semiconductor manufacturing information
JP4654939B2 (en) Color filter manufacturing plant monitoring system
Ison et al. Robust fault detection and fault classification of semiconductor manufacturing equipment
CN109063218A (en) A kind of control method and system of statistic processes
JP3750220B2 (en) Manufacturing method of semiconductor device
JP2004296826A (en) Quality control system and quality control method