TWI397830B - Versatile semiconductor manufacturing controller with statistically repeatable response times - Google Patents
Versatile semiconductor manufacturing controller with statistically repeatable response times Download PDFInfo
- Publication number
- TWI397830B TWI397830B TW95115310A TW95115310A TWI397830B TW I397830 B TWI397830 B TW I397830B TW 95115310 A TW95115310 A TW 95115310A TW 95115310 A TW95115310 A TW 95115310A TW I397830 B TWI397830 B TW I397830B
- Authority
- TW
- Taiwan
- Prior art keywords
- program
- controller
- control
- statistically
- short interval
- Prior art date
Links
Landscapes
- General Factory Administration (AREA)
- Programmable Controllers (AREA)
Description
此申請書主張由發明人雷諾羅森布(Lenoid Rozenboim)以及大衛高區(David Gosch)於2005年5月2日申請之名稱為“具有統計上可重複之回應時間的通用半導體製造控制器”之美國臨時專利申請案第60/676,770號。此臨時申請案以參考方式包含於此。This application claims to be a general-purpose semiconductor manufacturing controller with a statistically repeatable response time, which was invented by the inventor Lenoid Rozenboim and David Gosch on May 2, 2005. U.S. Provisional Patent Application No. 60/676,770. This provisional application is hereby incorporated by reference.
此申請案有關於共同擁有之由發明人烏茲理阿米(Uzi Lev-Ami)以及約瑟夫依蘭瑞奇(Yossef Ilan Reich)於2001年8月22日申請之名稱為“監視主機至工具通訊之方法與系統”之美國專利申請案第09/935,213號;由發明人烏茲理阿米(Uzi Lev-Ami)、關特斯夫納許(Guenter Sifnatsch)以及馬克愛特伍(Mark Attwood)於2004年4月7日申請之名稱為“針對半導體廠應用傳達來自智慧感應器之資料收集之控制器與方法”之美國專利申請案第10/819,903號以及美國專利申請案第10/948,085號。那些共同申請之申請書以參照方式包含於此如同其完全在此提出。This application has been filed on August 22, 2001 by Uzi Lev-Ami and Yossef Ilan Reich, jointly owned by the inventor, for the purpose of monitoring the host-to-tool communication. U.S. Patent Application Serial No. 09/935,213, the disclosure of which is incorporated by the inventors Uzi Lev-Ami, Guenter Sifnatsch, and Mark Attwood. U.S. Patent Application Serial No. 10/819,903, filed on Apr. 7, 2004, entitled <RTI ID=0.0>> . Applications for co-applications are hereby incorporated by reference as if they were fully incorporated herein.
本發明係有關具有統計上可重複之回應時間的通用半導體製造控制器。The present invention is directed to a general purpose semiconductor fabrication controller having a statistically repeatable response time.
本發明有關於半導體製造用之程序I/O控制器,工具主機可委派資料收集、監視與控制的工作給該程序I/O控制器。特別係,本發明有關於可以統計上可重複之效能與精準度執行超過一個的資料收集、監視、控制以及回應來自工具主機之命令。所述之實施例使用具優先順序之即時操作系統來控制半導體製造工具以及從與感應器關聯的工具之資料的收集。於選定的製作步驟期間對選定的命令以及感應器輸入之統計上可重複的回應性能有效降低顫動。The invention relates to a program I/O controller for semiconductor manufacturing, and the tool host can delegate the work of data collection, monitoring and control to the program I/O controller. In particular, the present invention relates to the ability to perform more than one data collection, monitoring, control, and response from a tool host with statistically repeatable performance and accuracy. The described embodiments use a prioritized, immediate operating system to control the collection of semiconductor manufacturing tools and materials from tools associated with the sensors. Vibrating is effectively reduced for the selected command and the statistically repeatable response performance of the sensor input during the selected fabrication steps.
莫爾(Moore)定律保證計算能力呈指數成長且具有更便宜的價格。處理能力之動態成長可能會讓人覺得半導體裝置製造可能是個有利可圖的事業,如同石油探勘。事實正好相反。由於製造批次價值很高且製造程序會對很小的錯誤敏感,半導體裝置製造係非常謹慎的事業。新的設備之認證週期以及標準以及舊有設備的修改冗長且繁雜。在釋出量產前,即使很小的改變也必須很詳盡的檢查。Moore's law guarantees that computing power grows exponentially and has a cheaper price. The dynamic growth of processing power may make people think that semiconductor device manufacturing may be a profitable business, like oil exploration. The opposite is true. Semiconductor device manufacturing is a very cautious business because of the high value of manufacturing batches and the manufacturing process that is sensitive to small errors. The certification cycle for new equipment, as well as standards and modifications to legacy equipment, are lengthy and cumbersome. Even small changes must be thoroughly examined before mass production is released.
半導體廠所使用的關鍵零件包含工具(如沉積室、反應爐)、監視工具之感應器(如FTIR感應器、質量攝譜儀、熱偶)以及儲存並分析來自感應器有關工具操作之資料的主機或分布之處理器。Key components used in semiconductor plants include tools (such as deposition chambers, reactors), sensors for monitoring tools (such as FTIR sensors, mass spectrographs, thermocouples), and storage and analysis of information from sensors related to tool operation. Host or distributed processor.
一先前申請案描述使用諸如於乙太網路上之TCP/IP的高速與抗錯誤之技術聆聽來自感應器之資料將之提供給主機或分布的處理器之透明的方法。此先前申請案係由發明人烏茲理阿米(Uzi Lev-Ami)以及約瑟夫依蘭瑞奇(Yossef Ilan Reich)於2001年8月22日申請之名稱為“監視主機至工具通訊之方法與系統”之美國專利申請案第09/935,213號,其以參考方式包含於此。該先前申請案描述可使用光學性隔離之連接器竊聽來自工具或感應器之序列通訊之聆聽埠。使用竊聽的方式,證明可無須工具或感應器之修改以低風險更新半導體廠通訊以及資料收集架構。可示範更新之可能性而無須拆解現有的通訊架構。A prior application describes a transparent method of listening to data from a sensor to provide a host or distributed processor using high speed and error resistant techniques such as TCP/IP over Ethernet. This prior application was filed on August 22, 2001 by the inventors Uzi Lev-Ami and Yossef Ilan Reich as the method of monitoring host-to-tool communication. U.S. Patent Application Serial No. 09/935,213, the disclosure of which is incorporated herein by reference. This prior application describes the use of optically isolated connectors to eavesdrop on the listening communication from the serial communication of the tool or sensor. The use of eavesdropping proves that semiconductor plant communications and data collection architectures can be updated with low risk without the need for tools or sensor modifications. The possibility of updating can be demonstrated without the need to disassemble the existing communication architecture.
下一代半導體廠儀器以及後端分析能力將涉及增加智慧控制器,如程序I/O控制器,以在程序I/O控制器之一側上,在工具與感應器之間傳達通訊,以及在另一側上,工具主機或分布的處理器,而無須替換或感變感應器之分析特性。增加的處理器能力以及降低的儲存成本出現先前於半導體廠中無法實現之組態的機會。由發明人烏茲理阿米(Uzi Lev-Ami)、關特斯夫納許(Guenter Sifnatsch)以及馬克愛特伍(Mark Attwood)於2004年4月7日申請之名稱為“針對半導體廠應用傳達來自智慧感應器之資料收集之控制器與方法”之美國專利申請案第10/819,903號描述具有各種能力之智慧型控制器。上述智慧型控制器缺少的係與工具主機配合同時進行多項功能且同時提供統計上可重複的回應性之能力。命令開始或完成時之顫動於目前軟體結構中並無法良好的掌控。Next-generation semiconductor plant instrumentation and back-end analysis capabilities will involve adding smart controllers, such as program I/O controllers, to communicate between tools and sensors on one side of the program I/O controller, and On the other side, the tool host or distributed processor does not require replacement or sensing characteristics of the sensor. Increased processor power and reduced storage costs present opportunities for configurations that were previously unachievable in semiconductor plants. On April 7, 2004, the inventors Uzi Lev-Ami, Guenter Sifnatsch, and Mark Attwood applied for the application for the semiconductor factory. A smart controller having various capabilities is described in U.S. Patent Application Serial No. 10/819,903, which is incorporated herein by reference. The lack of the above-mentioned smart controllers cooperates with the tool host to perform multiple functions while providing statistically repeatable responsiveness. The jitter at the beginning or completion of the command is not well controlled in the current software structure.
藉由委派來自工具主機或分布的處理器之資料收集與關鍵控制給程序I/O控制器,出現改變應用至處理室之控制模型的機會。更佳地,可導致具有統計上可重複的回應性之更簡單組態與控制以及更有彈性之零件與系統。By delegating data collection and critical control from the tool host or distributed processor to the program I/O controller, there is an opportunity to change the control model applied to the processing room. More preferably, it results in simpler configuration and control with statistically repeatable responsiveness and more flexible parts and systems.
本發明有關於半導體製造用之程序I/O控制器,工具主機可委派資料收集、監視與控制的工作給該程序I/O控制器。特別係,本發明有關於可以統計上可重複之效能與精準度執行超過一個的資料收集、監視、控制以及回應來自工具主機之命令。所述之實施例使用具優先順序之即時操作系統來控制半導體製造工具以及從與感應器關聯的工具之資料的收集。於選定的製作步驟期間對選定的命令以及感應器輸入之統計上可重複的回應性能有效降低顫動。本發明之特定的態樣係描述於申請專利範圍、說明書以及圖示中。The invention relates to a program I/O controller for semiconductor manufacturing, and the tool host can delegate the work of data collection, monitoring and control to the program I/O controller. In particular, the present invention relates to the ability to perform more than one data collection, monitoring, control, and response from a tool host with statistically repeatable performance and accuracy. The described embodiments use a prioritized, immediate operating system to control the collection of semiconductor manufacturing tools and materials from tools associated with the sensors. Vibrating is effectively reduced for the selected command and the statistically repeatable response performance of the sensor input during the selected fabrication steps. The specific aspects of the invention are described in the scope of the claims, the description and the drawings.
參照圖示進行下列詳細說明。描述較佳之實施例以說明本發明,而不限制本發明之範圍,其係由申請專利範圍所界定。熟悉該項技藝者可從下列說明理解到各種等效變異。The following detailed description is made with reference to the drawings. The preferred embodiments are described to illustrate the invention without limiting the scope of the invention, which is defined by the scope of the claims. Those skilled in the art will appreciate various equivalent variations from the following description.
從歷史角度來看,於半導體廠中,主機系統在大型電腦(main frame)、迷你電腦或工作站上運作。主機系統典型為單石構造的(monolithic),控制並監視半導體廠中的所有或一大組的工具。主機系統依賴轉接器以與工具和感應器接介。主機系統典型自工具與感應器接收資料並發送控制指令至工具。主機系統常接收並產生大量的序列通訊信息。Historically, in semiconductor factories, host systems operate on mainframes, minicomputers, or workstations. The host system is typically monolithic, controlling and monitoring all or a large set of tools in a semiconductor plant. The host system relies on an adapter to interface with tools and sensors. The host system typically receives data from the tool and sensor and sends control commands to the tool. The host system often receives and generates a large amount of serial communication information.
工具主機一詞廣義而言包含工具控制主機以及較限制或彈性分布之處理器。工具主機包含具有綜合積體工具控制功能之主機以及運行在分布的處理器上具有較有限且工作專門之功能的主機兩者。工具主機包含諸如Consilium的FAB300T M 軟體之產品,其係描述為提供由客戶特定之商業程序的集中定義所驅動之單一綜合工廠管理系統。此種類的工具主機係設計成取代傳統製造執行系統,其係設計成控制由不同廠商提供之工具。在工具主機系列中與傳統製造執行系統之相對側上,可在分布的處理器上運作零件程序以處置各種特定功能,稱不上綜合管理的系統。沿著該系列,諸如Consilium的FAB300T M 軟體之產品針對一些目的可被視為工具控制主機,而針對其他目的可被視為在分布的處理器上運作之程序。The term tool host broadly includes a tool control host and a more restrictive or elastically distributed processor. The tool host includes both a host with integrated integrator tool control and a host with limited functionality and job-specific functionality running on distributed processors. The tool host contains products such as Consilium's FAB300 T M software, which is described as providing a single integrated plant management system driven by a centralized definition of customer-specific business processes. This type of tool host is designed to replace traditional manufacturing execution systems designed to control tools provided by different vendors. On the opposite side of the tool host family from the traditional manufacturing execution system, the part program can be run on a distributed processor to handle a variety of specific functions, not an integrated management system. Along the series, products such as Consilium's FAB300 T M software can be considered a tool control host for some purposes, and can be viewed as a program running on a distributed processor for other purposes.
在此揭露之方法使用程序I/O控制器。程序I/O控制器發送並接收控制一處理室或其他半導體製造處理裝置的信號。The method disclosed herein uses a program I/O controller. The program I/O controller sends and receives signals that control a processing chamber or other semiconductor manufacturing processing device.
程序I/O控制器與資料使用者通訊從感應器收集而來之資料。資料使用者可為在大型電腦上運作之傳統工具主機或可為在分布的處理器上運作之較新的軟體。資料使用者可為單石系統或獨立或一起配合操作之聯合封裝。程序I/O控制器亦可監視來自感應器之資料、辨別關注的事件並對已經收集之資料請求更多資料或改變感應器之收集計畫,以回應受監視之資料。The program I/O controller communicates with the data user to collect information from the sensor. Data users can be traditional tool hosts that operate on large computers or newer software that can run on distributed processors. The data user can be a monolithic system or a joint package that operates independently or together. The program I/O controller can also monitor the data from the sensor, identify the event of interest and request more information from the collected data or change the sensor's collection plan in response to the monitored data.
第1圖描述於其中本發明之態樣特別有用之環境。此描述處理室125、來自處理室加上感應器、控制通道與控制器之各種輸入與輸出。室125可用於各種反應作用,如沉積、清洗、蝕刻、植入、灰化等等。其他工具的種類,未圖示,亦可從本發明之態樣得益。Figure 1 depicts an environment in which the aspects of the invention are particularly useful. This describes the processing chamber 125, various inputs and outputs from the processing chamber plus sensors, control channels, and controllers. Chamber 125 can be used for various reactions such as deposition, cleaning, etching, implantation, ashing, and the like. Other types of tools, not shown, may also benefit from the aspects of the present invention.
潛在可透過網際網路、虛擬私人網路或廣域網路112存取之半導體廠網路111具有經由控制器、防火牆或其他連接器162至工具網路112之受控的存取。於此圖中工具網路係顯示為連接影響處理室125之控制與感應器至一圓環(ring)。熟悉該項技藝者將了解到此架構僅為例示性;於半導體廠中比圓環更可能使用序列通訊、乙太網路或階層式通訊。The semiconductor plant network 111 potentially accessible via the Internet, virtual private network or wide area network 112 has controlled access via a controller, firewall or other connector 162 to the tool network 112. The tool network in this figure is shown as connecting the control and sensor that affects the processing chamber 125 to a ring. Those skilled in the art will appreciate that this architecture is merely illustrative; it is more likely to use serial communication, Ethernet or hierarchical communication than a ring in a semiconductor factory.
至反應室125之氣體輸入包含通過氣體箱壓力變換器113以及質量流控制器(MFC)114之氣體。某些氣體可通過臭氧產生器133。其他氣體或氣體混合物可通過反應性氣體產生器115以及氣體成分監視器117。反應性氣體產生器115可產生電漿,於處理室125之外或內。氣體成分監視器117可與反應性氣體產生器串聯或並聯。質量流控制器114可與反應性氣體產生器115以及氣體成分監視器117氣體上通訊,並最終或直接與處理室125氣體上通訊。氣體輸入裝置113、114、133、115以及117與一或更多數位控制器142、室控制器152以及連接點162通訊。此通訊通常包含控制與遙測技術兩者。這些裝置可包含回應裝置或氣體輸入以及/或輸出之操作的控制以及感應器兩者。The gas input to the reaction chamber 125 contains gas passing through the gas box pressure transducer 113 and the mass flow controller (MFC) 114. Certain gases may pass through the ozone generator 133. Other gases or gas mixtures may pass through the reactive gas generator 115 and the gas composition monitor 117. The reactive gas generator 115 can generate plasma outside or within the processing chamber 125. The gas composition monitor 117 can be connected in series or in parallel with the reactive gas generator. The mass flow controller 114 can be in gas communication with the reactive gas generator 115 and the gas composition monitor 117 and ultimately or directly communicate with the process chamber 125 gas. Gas input devices 113, 114, 133, 115, and 117 are in communication with one or more bit controllers 142, chamber controller 152, and connection points 162. This communication typically includes both control and telemetry techniques. These devices may include both control of the operation of the response device or gas input and/or output, as well as sensors.
其它輸入可包含材料遞送134、冷卻子系統145以及各種電源注射器153、154與155。反應室125可為沉積室、蝕刻器、熱處理器或其他種類的反應器。它可為連接至多個反應器之自動處置的單元。取決於反應室之種類,材料遞送系統134可供應例如於工作件136上供沉積之材料。冷卻子系統145可幫忙調節室125內之溫度,由於大部分化學反應係以溫度敏感之速率進行。供應至室之電源可係微瓦電源153、用於產生電漿之RF電源154以及用於產生電漿並加熱室或供應至室之氣體或其他材料之DC電源155。其它輸入,如氣體輸入,與一或更多數位控制器142、室控制器152以及連接點162通訊。此通訊通常包含控制與遙測技術兩者。這些裝置可包含回應裝置之操作的控制或感應氣體輸入以及/或輸出兩者的控制以及感應器兩者。Other inputs may include material delivery 134, cooling subsystem 145, and various power injectors 153, 154, and 155. Reaction chamber 125 can be a deposition chamber, an etcher, a thermal processor, or other type of reactor. It can be a unit that is connected to multiple reactors for automatic disposal. Depending on the type of reaction chamber, material delivery system 134 can supply materials for deposition, such as on workpiece 136. The cooling subsystem 145 can help regulate the temperature within the chamber 125 since most of the chemical reactions are carried out at a temperature sensitive rate. The power source supplied to the chamber may be a microwatt power source 153, an RF power source 154 for generating plasma, and a DC power source 155 for generating plasma and heating the chamber or gas or other material supplied to the chamber. Other inputs, such as gas inputs, communicate with one or more of the bit controller 142, the chamber controller 152, and the connection point 162. This communication typically includes both control and telemetry techniques. These devices may include both control of the operation of the response device or control of both the inductive gas input and/or output and both of the sensors.
感應器可回應室的狀況或對來自室之排出物作動。回應室的狀況之感應器包含晶圓監視器116,其經由窗戶126看進去室125內以看查薄膜的厚度、圖案以及其他特性(如EPI-OnlineT M )、諸如用於蝕刻處理控制的具有干擾過濾器或干涉計之光學放射監視器之處理監視器127以及壓力變換器137。對來自室125之排出物作動之感應器包含洩漏偵測器146、真空計157以及排出監視器158。這些感應器可與壓力控制器148以及控制閥147互動,並且與真空零件以及/或子系統156互動。它們亦可與泵以及/或排出氣體清潔器互動,其未出現在圖中。這些感應器與一或更多數位控制器142、室控制器152以及連接點162通訊。此通訊通常包含控制與遙測技術兩者。與感應器通訊(如147、148以及156)之這些裝置可包含控制以及感應器兩者。The sensor can respond to the condition of the chamber or actuate the effluent from the chamber. The sensor that responds to the condition of the chamber includes a wafer monitor 116 that is viewed through the window 126 into the chamber 125 for viewing the thickness, pattern, and other characteristics of the film (e.g., EPI-Online T M ), such as for etching process control. A processing monitor 127 and a pressure transducer 137 having an optical radiation monitor that interferes with the filter or interferometer. The sensor that operates the effluent from chamber 125 includes a leak detector 146, a vacuum gauge 157, and a discharge monitor 158. These sensors can interact with the pressure controller 148 and the control valve 147 and interact with the vacuum components and/or subsystem 156. They can also interact with the pump and/or the exhaust gas cleaner, which does not appear in the figure. These sensors communicate with one or more of the bit controller 142, the chamber controller 152, and the connection point 162. This communication typically includes both control and telemetry techniques. These devices that communicate with the sensors (e.g., 147, 148, and 156) can include both control and sensors.
一監視與控制系統之希望的特性為它的確定性(determinism),亦即,每一個命令可在指定截止期限內開始或完成之把握。開始與完成兩者皆納入考量,因為一些命令需要大量的時間來完成,如五分鐘收集各五毫秒之取樣。確定性之資格係視為命令在指定回應時間內開始或完成的機率。為了測試確定性,演練至少一百萬個命令,並將它們的實際回應時間整理成直方表。希望至少99%的命令回應時間落在要求的期限內。此外,對於透過專用網路在專用測驗起始器(運作控制部分的電腦)上持續運作數小時之測驗,希望命令在指定(如5、3或1毫秒)時間內執行的機率盡可能達到更多的“九”,但至少99.99%,亦即可允許10,000命令中有一個延遲超過指定回應時間或完全不執行。The desired characteristic of a monitoring and control system is its determinism, that is, each command can be started or completed within a specified deadline. Both start and finish are taken into account, as some commands require a lot of time to complete, such as collecting five milliseconds of sampling in five minutes. Deterministic qualifications are considered to be the probability that an order will begin or complete within a specified response time. To test certainty, drill at least one million commands and organize their actual response times into a histogram. It is hoped that at least 99% of the order response time will fall within the required time limit. In addition, for a continuous operation of a few hours on a dedicated test starter (computer running the control section) through a dedicated network, it is desirable that the command be executed as much as possible within a specified (eg 5, 3 or 1 millisecond) time. The more "nine", but at least 99.99%, can also allow one of the 10,000 commands to have a delay that exceeds the specified response time or not execute at all.
任何通訊技術具有固有的惡化或否則無法傳遞數位資訊到其目的地之能力。這係因為於類比(電子)與數位領域兩者中之一些現象。因此,100%之確定性係無法達成的理想,而實際實施則係根據它們與理想多接近來評估。這些標準可以所謂的“九”或三西格馬(sigma)或五sigma效能來表示。Sigma係指標準差,其為統計上的測量。Any communication technology has the inherent ability to deteriorate or otherwise fail to deliver digital information to its destination. This is due to some of the phenomena in both analog (electronic) and digital fields. Therefore, 100% certainty is an ideal that cannot be achieved, and actual implementation is evaluated based on how close they are to the ideal. These criteria can be expressed in terms of so-called "nine" or three sigma or five sigma efficiencies. Sigma refers to the standard deviation, which is a statistical measure.
第2A-B圖將效能與變異統計描述為分布。單尾分布由曲線210描述。設定諸如三(3σ)或五(5σ)sigma或α= .01(兩個九)或α= .0001(四個九)之效能標準,其由截斷線211顯示。回應時間的分布符合標準若,例如99%的實際回應時間落在截斷線211的左邊。雙尾分布由曲線220描述。設定諸如三(3σ)或五(5σ)sigma或α= .01(兩個九)或α= .0001(四個九)之容限標準,其由在目標時間223兩翼側之截斷線222、224顯示。從目標時間之變異之分布符合標準若例如實際與目標時間之間的差的99%落在截斷線222、224之間。雖然所示之分布為常態分布,可應用標準統計或無母數標準來測量包含歪斜分布之其他分布。這些標準係施加至間隔夠遠的激源(stimulus)以不使系統超載。例如,程序I/O控制器之卡片機架實施例已測量為針對每一I/O命令在5毫秒內執行,當以至少20毫秒之間隔接收此種命令。此回應時間的限制因素為用於卡片機架座之中的序列CANbus底板以及在受支援的I/O卡中內部使用之其他序列協定。可預見到客制化之積體處理裝置將針對每一個I/O命令在少於3毫秒或1毫秒內執行這些命令,而不受限於命令的頻率。Figure 2A-B depicts the performance and variation statistics as a distribution. The one-tailed distribution is depicted by curve 210. A performance criterion such as three (3σ) or five (5σ) sigma or α = .01 (two nines) or α = .0001 (four nines) is set, which is indicated by the cut line 211. The distribution of the response time conforms to the standard. For example, 99% of the actual response time falls to the left of the cut line 211. The two-tailed distribution is depicted by curve 220. A tolerance criterion such as three (3σ) or five (5σ) sigma or α = .01 (two nines) or α = .0001 (four nines) is set, which is determined by the cut line 222 on both wings of the target time 223, 224 shows. The distribution from the variation of the target time conforms to the standard if, for example, 99% of the difference between the actual and target time falls between the cut lines 222, 224. Although the distribution shown is a normal distribution, standard statistics or no parent number criteria can be applied to measure other distributions including skew distributions. These criteria are applied to a stimulus that is far enough apart to not overload the system. For example, a card rack embodiment of a program I/O controller has been measured to execute within 5 milliseconds for each I/O command, when such commands are received at intervals of at least 20 milliseconds. The limiting factor for this response time is the sequence of CANbus backplanes used in card rack mounts and other serial protocols used internally in supported I/O cards. It is foreseen that the customized integrated processing device will execute these commands in less than 3 milliseconds or 1 millisecond for each I/O command, without being limited by the frequency of the commands.
根據經驗,已確定使用卡片機架座可降低顫動並使效能增加到符合5毫秒之標準。欲符合更確切之容限,諸如3毫秒或1毫秒,需要如下所述之更積體化以及客制化之硬體組態。Based on experience, it has been determined that the use of a card rack mount reduces chattering and increases performance to a standard of 5 milliseconds. To meet a more precise tolerance, such as 3 milliseconds or 1 millisecond, a more integrated and customized hardware configuration as described below is required.
第3圖為與工具、感應器以及工具主機或中央主控制器耦合之通訊協調器之方塊圖。此通訊協調器組態包含兩個SEC/GEM介面埠312與316以及兩個網路介面埠332與336。控制器包含邏輯與資源以經由包含SECS多工器(MUX)315之SECS協定通訊。它進一步包含實施半導體廠側介面334之邏輯與資源以與資料使用者通訊以及工具側介面335以與工具、感應器與儀器通訊。SECS MUX 315以及介面334與335邏輯上連接至資料收集與發表資源325。於通訊協調器之半導體廠側上,傳統的工具主機311可經由符合SECS之通訊通道312連接,其可為SECS-I、HSMS或SECS後來的版本或後續。於可應用本發明之非半導體廠之環境中,可使用其他協定來與諸如醫療工具或數字控制機器工具之受到監視之工具連接。它亦可經由與SECS不同之半導體廠側協定連接,由網路322承載至貯存331以及報告321資源。於通訊協調器之工具與感應器側上,可連接有至工具或工具群之SEC/GEM工具介面317。雖於此圖中顯示控制器透過網路連接至感應器,它亦可替代地使用SECS-I或另一以序列為基礎之協定,以連接至感應器。引進第3圖,可引導程序I/O控制器以處置工具介面316與儀器介面336兩者。程序I/O控制器可代替通訊協調器。Figure 3 is a block diagram of a communication coordinator coupled to a tool, sensor, and tool host or central host controller. This communication coordinator configuration includes two SEC/GEM interfaces 埠 312 and 316 and two network interfaces 埠 332 and 336. The controller contains logic and resources to communicate via the SECS protocol including the SECS multiplexer (MUX) 315. It further includes the logic and resources to implement the semiconductor factory side interface 334 to communicate with the data user and the tool side interface 335 to communicate with the tool, sensor, and instrument. The SECS MUX 315 and interfaces 334 and 335 are logically coupled to the data collection and publication resource 325. On the semiconductor plant side of the communication coordinator, the conventional tool host 311 can be connected via an SECS compliant communication channel 312, which can be a later version or subsequent to SECS-I, HSMS or SECS. In an environment where non-semiconductor plants to which the present invention can be applied, other protocols can be used to interface with monitored tools such as medical tools or digitally controlled machine tools. It can also be connected by a semiconductor factory side agreement different from SECS, carried by network 322 to storage 331 and reporting 321 resources. On the tool and sensor side of the communication coordinator, an SEC/GEM tool interface 317 to the tool or tool group can be connected. Although the controller is shown in the figure to be connected to the sensor via a network, it may alternatively use SECS-I or another sequence-based protocol to connect to the sensor. Introducing FIG. 3, the program I/O controller can be directed to handle both the tool interface 316 and the instrument interface 336. The program I/O controller can replace the communication coordinator.
第4圖為使用單一種類的通訊通道以與工具、感應器以及工具主機通訊之通訊協調器之方塊圖。第5圖描述多種通訊通道之使用。於第4圖中,描述SECS主導之狀況。通訊協調器403使用SECS協定407與409以與工具401以及感應器402通訊。工具402與感應器402之間的連結408涉及可正常感應到之任何形式之能量或力量,包含於第1圖之討論中辨別之任何知覺。於第5圖中,描述SECS與網路通訊之更複雜的結合。於此情況中唯一的SECS通訊通道407係介於通訊協調器403以及主機401之間。工具側網路509將通訊協調器與感應器402A-B、管理的交換器513以及網路附接儲存器(NAS)515連接在一起。半導體廠側網路519將通訊協調器與諸如傳統的工具主機或分佈的處理器之分析軟體523、諸如網際網路、VPN或專用內部網路之延伸網路525以及存留或儲存由通訊協調器發表之資料的資料庫527。替代地,資料庫可位在通訊協調器403上。將於下一章節討論替代的實際實施。於第4圖中,可將程序I/O控制器引進到通訊協調器403以及工具與感應器407與409之間,讓通訊協調器處置SECS協定信息。於第5圖中,可將程序I/O控制器引進到通訊協調器403以及工具與感應器401與402之間。替代地,除去SECS協定、管理的交換器513與NAS 515,程序I/O控制器可代替通訊協調器403。Figure 4 is a block diagram of a communication coordinator that uses a single type of communication channel to communicate with tools, sensors, and tool hosts. Figure 5 depicts the use of various communication channels. In Figure 4, the status of the SECS is described. The communication coordinator 403 uses the SECS protocols 407 and 409 to communicate with the tool 401 and the sensor 402. The bond 408 between the tool 402 and the sensor 402 relates to any form of energy or force that can be sensed normally, including any perception discerned in the discussion of FIG. In Figure 5, a more complex combination of SECS and network communication is described. The only SECS communication channel 407 in this case is between the communication coordinator 403 and the host 401. The tool side network 509 connects the communication coordinator with the sensors 402A-B, the managed switch 513, and the network attached storage (NAS) 515. The semiconductor factory side network 519 integrates the communication coordinator with an analysis software 523 such as a conventional tool host or distributed processor, an extension network 525 such as an internet, VPN or private internal network, and stores or stores the communication coordinator. A database of published materials 527. Alternatively, the database can be located on the communication coordinator 403. The actual implementation of the alternative will be discussed in the next section. In Figure 4, a program I/O controller can be introduced between the communication coordinator 403 and the tools and sensors 407 and 409 to allow the communication coordinator to process the SECS protocol information. In FIG. 5, a program I/O controller can be introduced between the communication coordinator 403 and the tools and sensors 401 and 402. Alternatively, instead of the SECS protocol, managed switch 513 and NAS 515, the program I/O controller can replace the communication coordinator 403.
第一個程序I/O控制器之實施係根據用於之前的產品中之3U“歐洲卡(Euro card)”之型態。它再利用現有的離散與類比I/O卡,其已經修改以使用在CANbus上之簡化之高效能協定。添加額外的一張3U卡,其實施兩個乙太網路埠(與第2層交換器組織互連),並支援程序I/O控制器軟體元件。於一實施例中,它使用於CANbus上之專屬(proprietary)協定以存取實際的I/O點。此實施為模組式,並於單一卡機架中支援多達16張I/O卡。系統於每一個可用之卡槽中自動偵測已插入哪一張I/O卡,並組態所有的軟體元件(控制與資料收集)以反映自動偵測到之組態。The implementation of the first program I/O controller is based on the 3U "Euro card" type used in the previous product. It reuses existing discrete and analog I/O cards that have been modified to use simplified high-performance protocols on CANbus. Add an additional 3U card that implements two Ethernet ports (interconnected with Layer 2 switches) and supports program I/O controller software components. In one embodiment, it uses a proprietary protocol on CANbus to access the actual I/O points. This implementation is modular and supports up to 16 I/O cards in a single card rack. The system automatically detects which I/O card has been inserted in each available card slot and configures all software components (control and data collection) to reflect the automatically detected configuration.
第6圖為程序I/O控制器實際架構的高度方塊圖。於此實施例中,SCAN匯流排621將卡機架中之卡連接至介面控制器631。介面控制器可為具有Motorola Coldfire 32位元處理器之Netcom卡、網路連接性(前端642與後端643)、一或更多具有RS232以及RS485支援的軟體選擇之UART以及熱插拔I/O用之內部CAN匯流排控制器。乙太網路連接器可在控制功能以及其他功能之間排列優先順序。卡片可包含連鎖卡611,如組態供雙插槽並具有36-68個繼電器與8-32個DIDO通道之MKS儀器CDN497-C-E。有效之連鎖可由連鎖卡上之狀態燈所表示。它們可包含數位I/O 612以及類比I/O 613卡,如具有48個進/出埠以低主動信號在24v操作之MKS儀器CDN497-C-E以及耦合至12位元的反相器之具有32個類比輸入與16個類比輸出之MKS儀器496-C-E。Figure 6 is a block diagram of the height of the actual architecture of the program I/O controller. In this embodiment, the SCAN bus 621 connects the card in the card bay to the interface controller 631. The interface controller can be a Netcom card with a Motorola Coldfire 32-bit processor, network connectivity (front-end 642 and back-end 643), one or more UARTs with RS232 and RS485 support, and hot-swap I/ O for the internal CAN bus controller. The Ethernet connector prioritizes control and other functions. The card may include an interlock card 611, such as the MKS instrument CDN497-C-E configured for two slots and having 36-68 relays and 8-32 DIDO channels. An effective chain can be represented by a status light on the chain card. They can include digital I/O 612 and analog I/O 613 cards, such as the MKS instrument CDN497-C-E with 48 in/out 埠 with low active signal operating at 24v and an inverter coupled to 12-bit. MKS instrument 496-C-E with 32 analog inputs and 16 analog outputs.
此模組化設計最適合小量之應用。它藉由再利用現有的I/O卡節省開發成本並僅需客制機座之開發以及針對各應用需求之分布板。在每單元的價格之基礎上,模組化設計比完全客制化的單石實施更昂貴。惟,它無法達到希望的快速回應時間與窄的容限。This modular design is best suited for small applications. It saves development costs by reusing existing I/O cards and requires only the development of custom bases and distribution boards for individual application needs. Modular design is more expensive than a fully customized single stone implementation based on the price per unit. However, it does not achieve the desired fast response time and narrow tolerance.
當以大量客制版本的方式提供遠端I/O單元時,其價格可更具競爭力。除去模組化可大幅降低硬體零件成本。從卡機架的實施例,學習到序列底板(基於CANbus)對效能有負面影響。有鑑於此,為了增進效能,可將所有I/O信號可透過平行底板與資料匯流排連接制處理器。When a remote I/O unit is provided in a large number of customized versions, the price can be more competitive. Eliminating modularization can significantly reduce the cost of hardware parts. From the embodiment of the card rack, learning that the sequence backplane (based on CANbus) has a negative impact on performance. In view of this, in order to improve performance, all I/O signals can be connected to the processor through a parallel backplane and a data bus.
硬體模組化(卡、槽、連接器)之排除無須邏輯之排除。藉由將所有I/O群集成為類似“邏輯卡”,可令主控制器韌體發現硬體中存在的I/O信號之種類與數量,並根據此組態執行所有軟體物件映照。此產生可運作不同硬體實施例之單一版本的軟體。在許多其他邏輯上類似但實際上不同的單元上可有新的特徵與除錯機制,得降低成本。The elimination of hardware modularization (cards, slots, connectors) does not require logical exclusion. By making all I/O clusters look like "logic cards", the main controller firmware can discover the type and number of I/O signals present in the hardware and perform all software object mappings based on this configuration. This produces a software that can operate a single version of a different hardware embodiment. There are new features and debugging mechanisms on many other logically similar but actually different units that reduce costs.
第7圖為具有MODBUS/TCP介面之I/O與COM架構之方塊圖。即時控制711與I/O驅動器731(亦即經由SCAN匯流排)、與MODBUS TCP驅動器733、資料記錄器712以及診斷器713耦合。I/O驅動器731可耦合至諸如CAN I/O通道之I/O通道741。可委派確定性操作之較高優先權給此圖中所示之工作,使用以搶先序(preemptive priority)為基礎之排序方式之即時嵌入式操作系統,並留下較低優先權給其他工作以及剩餘的資源之使用給較耗時但非關鍵之處理工作。工作713、724與734為具有沉重處理需求之較低優先權之工作。資料記錄器係耦合至資料收集723以及使用者介面724。資料收集以及使用者介面兩者可透過HTTP伺服器734在TCP/IP 743與乙太網路753上存取。資料記錄器712亦耦合至MODBUS TCP 733資源,其可透過閘道732與UART 742或透過TCP/IP 743與乙太網路753載送信息。Figure 7 is a block diagram of the I/O and COM architecture with MODBUS/TCP interface. The instant control 711 is coupled to the I/O driver 731 (i.e., via the SCAN bus), to the MODBUS TCP driver 733, the data logger 712, and the diagnostics 713. I/O driver 731 can be coupled to I/O channel 741, such as a CAN I/O channel. The higher priority of deterministic operations can be delegated to the work shown in this figure, using an instant embedded operating system based on a preemptive priority based ordering, leaving a lower priority for other work and The use of the remaining resources gives time-consuming but non-critical processing. Jobs 713, 724, and 734 are lower priority jobs with heavy processing requirements. The data logger is coupled to data collection 723 and user interface 724. Both the data collection and user interface can be accessed over TCP/IP 743 and Ethernet 753 via HTTP server 734. Data logger 712 is also coupled to MODBUS TCP 733 resources, which can carry information via gateway 732 and UART 742 or via TCP/IP 743 and Ethernet 753.
這兩種硬體實施例可使用現存標準MODBUS/TCP提供至I/O信號之存取。取而代之,可使用於較為精簡之UDP協定上之MODBUS的專屬實施,其中需要稍快之回應時間。選擇MODBUS可使程序I/O控制器快速且便宜地整合成新以及現有的半導體製造處理控制設備,因為MODBUS規格為開放式且可自由取得。可有數種MODBUS實施作為控制軟體整合之參考碼。Both hardware embodiments can provide access to I/O signals using existing standard MODBUS/TCP. Instead, it can be used for the exclusive implementation of MODBUS on a more streamlined UDP protocol, which requires a slightly faster response time. The choice of MODBUS allows the program I/O controller to be quickly and inexpensively integrated into new and existing semiconductor manufacturing process control equipment because the MODBUS specification is open and freely available. There are several MODBUS implementations that can be used as reference codes for controlling software integration.
第8圖為將程序I/O控制器820與830關聯至工具主機810之高度方塊圖。工具主機810包含控制程式811、MODBUS/TCP主應用程式介面812以及網路瀏覽器813。程序I/O控制器820包含MODBUS/TCP介面821以及網路伺服器822。MODBUS/TCP介面控制Javascript常規823之載入與執行並提供對I/O介面接腳825之直接的存取。網路伺服器822提供對資料收集824之存取。額外的程序I/O控制器830具有與820相同的能力。透過乙太網路815之通訊可將工具主機連接至程序I/O控制器或可將程序I/O控制器820與830互連。FIG. 8 is a height block diagram of the association of program I/O controllers 820 and 830 to tool host 810. The tool host 810 includes a control program 811, a MODBUS/TCP main application interface 812, and a web browser 813. The program I/O controller 820 includes a MODBUS/TCP interface 821 and a network server 822. The MODBUS/TCP interface controls the loading and execution of the Javascript regular 823 and provides direct access to the I/O interface pin 825. Network server 822 provides access to data collection 824. The additional program I/O controller 830 has the same capabilities as 820. The tool host can be connected to the program I/O controller via the communication of the Ethernet 815 or the program I/O controllers 820 and 830 can be interconnected.
程序I/O控制器實施例使用有效且確定性之嵌入式的即時軟體平台以有效地超越包含DeviceNet之傳統工業網路之效能與確定性,藉此降低涉及遷移控制系統之風險。下列為MODBUS遠端I/O控制之這些程序I/O控制器實施所提供之一些特徵。The Program I/O Controller embodiment uses an efficient and deterministic embedded instant software platform to effectively exceed the performance and certainty of traditional industrial networks including DeviceNet, thereby reducing the risk involved in migrating control systems. The following are some of the features provided by these program I/O controller implementations for MODBUS remote I/O control.
在最低層級,某些專門應用之繼電器電路或連鎖設計成防止處理機器進入到潛在危險的狀態。此方法傳統上已用作任何半導體處理控制應用的一部分,且為需要客制建立之控制特定I/O子系統設計之主要因素之一。這些連鎖可防止邏輯軟體錯誤或可能觸發氣體以及/或能量程度之潛在危險的組合之任何其他未預期之事件。一些程序I/O控制器實施例產品客制或半客制化之繼電器或連鎖電路,以及允許透過網路連結監視這些繼電器之額外的電子。At the lowest level, some specialized application relay circuits or interlocks are designed to prevent the processing machine from entering a potentially dangerous state. This method has traditionally been used as part of any semiconductor processing control application and is one of the main factors that require custom-built control of a particular I/O subsystem design. These linkages may prevent any other unexpected event of a logical software error or a combination of potential hazards that may trigger a gas and/or energy level. Some program I/O controller embodiments are custom or semi-customized relays or interlock circuits, as well as additional electronics that allow monitoring of these relays via a network connection.
在第二層級,控制器I/O裝置可內部防止其嵌入式之軟體的失敗。I/O信號可由可編程邏輯陣列(如CPLD)硬體控制,其可將所有離散輸出關閉,若程序I/O控制器韌體無法於預定時間間隔提供“心跳”脈衝制I/O電路。類似地,類比輸出將重設成零伏特位階,若韌體看似失敗。這些輸出的狀態,離散輸出之關閉狀態以及類比之0伏特,於半導體處理設備控制系統中傳統被指定為安全的輸出狀態。At the second level, the controller I/O device can internally prevent the failure of its embedded software. The I/O signals can be controlled by a programmable logic array (such as a CPLD) that can turn off all discrete outputs if the program I/O controller firmware is unable to provide a "heartbeat" pulsed I/O circuit at predetermined intervals. Similarly, the analog output will be reset to a zero volt level if the firmware appears to fail. The state of these outputs, the off state of the discrete outputs, and the analogy of 0 volts, have traditionally been designated as safe output states in semiconductor processing equipment control systems.
於最高層級,韌體特徵維持輸出現行以及維持心跳脈衝,只要程序I/O控制器持續以預定之規律的頻率發送MODBUS命令。若運行控制邏輯之主電腦失敗,程序I/O控制器韌體將立刻轉變所有輸出到安全狀態,直到工具主機恢復操作。可藉由切換操作模式到診斷而取消程序I/O控制器產品中的此特徵。每當選擇了診斷或任何較高層級之除錯操作模式時,程序I/O控制器單元可藉由在其前端面板上顯示黃或琥珀色之狀態燈(取代正常的綠色指示)來清楚地表示此模式。At the highest level, the firmware feature maintains the output current and maintains the heartbeat pulse as long as the program I/O controller continues to send MODBUS commands at a predetermined frequency. If the host computer running the control logic fails, the program I/O controller firmware will immediately transition all outputs to a safe state until the tool host resumes operation. This feature in the program I/O controller product can be eliminated by switching the operating mode to diagnostics. Whenever a diagnostic or any higher level debug mode of operation is selected, the program I/O controller unit can be clearly indicated by displaying a yellow or amber status light on its front panel (instead of the normal green indication). Indicates this mode.
這些程序I/O控制器呈現遠端I/O單元之網路介面。具有網際網路之瀏覽器之個人電腦可用來針對各種需求直接存取I/O單元使用者介面。此使用者網路介面係在由程序I/O控制器提供之正常MODBUS控制服務之外額外的,並可在處理設備的開發、製造與維持期間使用。在開發期間,嵌入式之網路使用者介面允許工程師建立並測試他們的處理設備而無須等到專門應用的軟體完全開發並經過測試。These program I/O controllers present the network interface of the remote I/O unit. A personal computer with an Internet browser can be used to directly access the I/O unit user interface for various needs. This user network interface is additional to the normal MODBUS control services provided by the program I/O controller and can be used during the development, manufacture, and maintenance of the processing device. During development, the embedded web user interface allows engineers to build and test their processing equipment without having to wait until the software for the application is fully developed and tested.
網路介面避免處理控制軟體需要暴露在低層級之診斷介面,並允許技術人員監視輸入並在最低層級運用輸出。可由網路診斷介面來滿足製造測試之需求。對於維護除錯-無須特別的軟體或硬體工具;個人電腦可用來透過乙太網路埠連接至I/O單元。並且,可大幅降低對於現場人員之訓練需要,因為診斷用之網路介面係非常憑直覺並可幾乎無須任何特別的訓練加以使用。但診斷有時會與安全需求衝突。因此,當I/O單元係在生產模式中時,診斷網路介面會降至僅監視之模式。The network interface avoids processing control software that needs to be exposed to lower-level diagnostic interfaces and allows technicians to monitor inputs and use outputs at the lowest level. The network diagnostic interface can be used to meet the needs of manufacturing testing. For maintenance debugging - no special software or hardware tools are required; PCs can be used to connect to I/O units via Ethernet. Moreover, the training needs for field personnel can be greatly reduced, because the diagnostic network interface is very intuitive and can be used with almost no special training. But the diagnosis sometimes conflicts with security requirements. Therefore, when the I/O unit is in production mode, the diagnostic network interface is reduced to the monitor-only mode.
當在生產模式中,程序I/O控制器韌體能致能一安全監視器(watchdog),其會使輸出維持有效只要工具主機為有效,但將維持僅工具主機或主或中央控制器對輸出狀態的專一之控制,並且亦將藉由於前端面板上閃爍綠色狀態指示燈而指示此狀態。但當切換程序I/O控制器韌體(如藉由一旋轉切換器)至診斷模式時,安全監視器將會無效,藉此移除工具主機之需要,並允許網路使用者介面對輸出之控制,並以琥珀色狀態燈之閃爍來指示此狀態。When in production mode, the program I/O controller firmware enables a security watchdog that keeps the output valid as long as the tool master is active but will maintain only the tool master or master or central controller output The state is controlled exclusively and will also be indicated by a flashing green status indicator on the front panel. However, when switching the program I/O controller firmware (such as by a rotary switch) to the diagnostic mode, the security monitor will be invalid, thereby removing the need of the tool host and allowing the network user to face the output. Control and indicate this state with a flashing amber status light.
程序I/O控制器產品架構之一部分係支援MKS儀器TOOLwebT M 架構以進行處理錯誤偵測與先進處理控制之資料收集與分析。TOOLweb後來的版本使用從盡可能多的處理變數收集而來之高品質資料。資料收集代理器之程序I/O控制器之實施,結合了TOOLweb之ToolSide(工具側)協定,令幾乎所有其之I/O信號供以高頻收集。因此,包含所述之特徵的程序I/O控制器替TOOLweb資料收集與基本處理之詳盡分析產生一個機會。Part of the program I/O controller product architecture supports the MKS instrument TOOLweb T M architecture for data collection and analysis of process error detection and advanced process control. Later versions of TOOLweb use high quality materials collected from as many processing variables as possible. The implementation of the data collection agent's program I/O controller, combined with TOOLweb's ToolSide protocol, allows almost all of its I/O signals to be collected at high frequencies. Therefore, the program I/O controller containing the described features creates an opportunity for detailed analysis of TOOLweb data collection and basic processing.
經驗證實目前處理控制軟體資料收集實施通常缺乏複雜數學模型方法需要之品質與總處理能力。目前的控制軟體係主要用在控制處理,因此非控制必要之任何項目傾向在實施者的優先順序表單中較低的位置,因此若實施,通常不足夠。Experience has confirmed that the current implementation of process control software data collection often lacks the quality and total processing power required for complex mathematical model methods. The current control soft system is mainly used for control processing, so any item that is not controlled tends to be in a lower position in the implementer's priority order, so if implemented, it is usually not enough.
監視半導體製程之感應器用來控制程序,並將某些關鍵處理參數維持在分配的容限內。此外,在晶圓完成後,越來越常收集、儲存並分析資料以用作缺點分析。收集並儲存來自連接至處理室之感應器之讀數以供離線分析,其旨在將各種資料資源互相關聯以發現處理失敗的根源,或預測處理設備上所需之防止性的維護。感應器通常包含各種氣體壓力感應器、溫度變換器、質量流控制器、RF與DC電源與能量傳送設備等等。Sensors that monitor semiconductor processes are used to control the program and maintain certain critical processing parameters within the assigned tolerances. In addition, after wafer completion, data is increasingly collected, stored, and analyzed for use as a defect analysis. The readings from the sensors connected to the processing chamber are collected and stored for off-line analysis, which is intended to correlate various data resources to discover the source of processing failures or to predict the preventive maintenance required on the processing equipment. Sensors typically include various gas pressure sensors, temperature transducers, mass flow controllers, RF and DC power and energy transfer devices, and more.
收集的資料應具備可供後續數學或統計分析之品質。測量設備應產生基礎之物理量之正確的數位讀取數。應在比任何處理量之任何預期的改變速率小許多之間隔取得測量取樣,因此可由使用的任何分析方法輕易的描述出改變速率或改變速率或圖案中之任何異常。應在指定與統計上可重複的間隔取得取樣。The information collected should be of a quality that can be used for subsequent mathematical or statistical analysis. The measuring device shall produce the correct number of digital reads of the underlying physical quantity. The measurement samples should be taken at intervals that are much smaller than any expected rate of change for any throughput, so any rate of change or rate of change or any anomaly in the pattern can be readily described by any analytical method used. Sampling should be taken at specified and statistically repeatable intervals.
當儀器取得每一個測量的當下應紀錄可從其導出取樣時間之值或至少取樣數字。收集資料之此時間戳可與其值同樣重要,特別係若取樣時間中有顫動。紀錄或導出之引信(fuse)應至少與測量值同樣正確,否則,將無法用來計算變量之導數。物理測量之導數反映改變速率或給定測量之“坡度”,並可在偵測各種處理異常中扮演重要的角色,或有效地預測防止性測量之需要。例如,在室加熱階段期間,溫度導數之逐漸下降(其中該溫度坡度為正的)代表加熱源之逐漸疲乏,並且此量之顯著的陡降則為使用之加熱源已經無效的肯定之徵兆。冷卻時,其中溫度導數為負的,相對於前次此量之下降則暗示冷卻可能太迅速,其會增加受處理之材料(亦即晶圓)中晶體結構壓力。When the instrument obtains the current record of each measurement, the value of the sampling time or at least the sampled number can be derived therefrom. This time stamp for collecting data can be as important as its value, especially if there is chattering during the sampling time. The fuze recorded or derived should be at least as good as the measured value, otherwise it will not be used to calculate the derivative of the variable. The derivative of the physical measurement reflects the rate of change or the "slope" of a given measurement and can play an important role in detecting various processing anomalies or effectively predicting the need for preventive measurements. For example, during the chamber heating phase, a gradual decrease in the temperature derivative (where the temperature gradient is positive) represents a gradual fatigue of the heating source, and a significant steep drop in this amount is a positive sign that the heating source used has been ineffective. When cooled, where the temperature derivative is negative, a decrease relative to the previous amount implies that the cooling may be too rapid, which increases the crystal structure pressure in the material being processed (ie, the wafer).
以先進資料收集為考量而開發之感應器與材料傳送控制器中,資源使用係嚴格地排列優先順序以使控制優先於資料收集,而不會負面影響收集到之資料的品質。此涉及具有優先排序之即時操作系統,該優先排序負責確保對控制命令以及測量詢問之統計上可重複之回應,同時讓資料收集工作與報告使用剩餘的資源。In sensor and material transfer controllers developed with advanced data collection considerations, resource usage is strictly prioritized to prioritize control over data collection without adversely affecting the quality of the data collected. This involves a prioritized, immediate operating system that is responsible for ensuring a statistically repeatable response to control commands and measurement queries while allowing the data collection work and reporting to use the remaining resources.
控制與資料收集使用不同的協定有其好處,這是因為這兩種活動之需求。控制著重於確定性之回應時間與有效性,而資料收集則喜好低複雜度以及與IT系統和資料庫之整合容易度。使用不同的協定亦區分出擴展這兩種活動之軟體元件。於某些情況中,甚至針對這些活動的每一個使用個別的硬體介面為恰當的,例如,針對控制使用DeviceNet或類比介面,以及針對資料收集使用IT友善之乙太網路埠以及TCP/IP與HTTP網路技術。於其他情況中,在用於HTTP/XML資料收集協定之相同的乙太網路埠上運作控制用之MODBUS/TCP也頗為恰當。The use of different agreements for control and data collection has its benefits because of the need for both activities. Control focuses on deterministic response time and effectiveness, while data collection prefers low complexity and ease of integration with IT systems and databases. The use of different protocols also distinguishes between software components that extend these two activities. In some cases, it is appropriate to use an individual hardware interface for each of these activities, for example, using DeviceNet or an analog interface for control, and using IT-friendly Ethernet and TCP/IP for data collection. With HTTP network technology. In other cases, MODBUS/TCP for operational control on the same Ethernet network for HTTP/XML data collection protocols is also appropriate.
資料收集的一種方法係令感應器備有裝置以確保資料收集之精準的定時。“資料收集計畫”係引進感應器之韌體內。資料收集客戶端因而需定義一個收集計畫,其將觸發輕型資料收集常駐程式以開始在精準的間隔取樣資料、增加時間戳至資料取樣並且儲存資料於預定尺寸的環形緩衝器中。One method of data collection is to have sensors equipped to ensure accurate timing of data collection. The “data collection plan” is the introduction of the body of the sensor. The data collection client thus needs to define a collection plan that will trigger the lightweight data collection resident program to begin sampling data at precise intervals, adding timestamps to data samples, and storing the data in a circular buffer of predetermined size.
此後,資料收集客戶端將產生資源優先權允許之週期性的請求,以擷取有關於特定收集計畫之經緩衝的取樣,並典型地接收自從上一個資料傳輸紀錄的所有的取樣。資料擷取請求之間隔無須準確或為確定性的,只要請求在避免環形緩衝器超載之間隔進行。由於資料附有時間戳,任何資料擷取的潛伏將不會對收集到的資料有影響。若緩衝器夠大能保留數秒的取樣,則幾乎可保證不會資料遺失,除非有資料收集客戶端或基礎網路架構之毀滅性的失敗。於收集計畫的界定中,資料收集客戶端界定將收集哪些測量,且多常。具有足夠之資源的感應器可容納多個資料客戶端與收集計畫。Thereafter, the data collection client will generate a periodic request for resource priority to retrieve buffered samples for a particular collection plan and typically receive all samples from the previous data transmission record. The interval between data retrieval requests need not be accurate or deterministic as long as the request is made at intervals that avoid ring buffer overload. Due to the time stamp attached to the data, the latency of any information will not affect the information collected. If the buffer is large enough to hold a few seconds of sampling, there is almost no data loss unless there is a devastating failure of the data collection client or the underlying network architecture. In the definition of the collection plan, the data collection client defines which measurements will be collected and how often. Sensors with sufficient resources can accommodate multiple data clients and collection plans.
資料取樣可隨意地加上序列數。資料收集客戶端使用序列數以避免紀錄重複之資料取樣並偵測遺失之資料取樣。序列數一支援資料收集伺服器之無狀態的實施,允許資料收集伺服器追蹤上一個擷取之資料取樣之序列數,並請求以下一個序列數為起頭之資料。已知無狀態伺服器實施對記憶體與CPU資源兩者的使用很有效率,降低感應器儲存資料之資源要求。Data sampling can be arbitrarily added to the sequence number. The data collection client uses the sequence number to avoid recording duplicate data samples and detecting missing data samples. The sequence number 1 supports the stateless implementation of the data collection server, allowing the data collection server to track the sequence number of the last captured data sample and request the following sequence number as the starting data. It is known that stateless servers implement efficient use of both memory and CPU resources, reducing the resource requirements for sensor storage data.
可實質上以批次處理對感應器緩衝器之資料取樣的擷取,使得客戶端資料請求服務多個資料取樣,多達緩衝器之最大數量。這會擴展對多個資料取樣之資料擷取請求中關聯之酬載,使資料收集處理更有效運用頻寬,並降低產生之平均酬載。The sampling of the data of the sensor buffer can be processed substantially in batches, so that the client data request services multiple data samples, up to the maximum number of buffers. This will expand the data associated with multiple data sampling requests, so that the data collection process can more effectively use the bandwidth and reduce the average payload generated.
於各種開發階段中有數個感應器與材料傳送產品實施取樣緩衝。它們使用應用XML之資料模型,其係於HTTP之無狀態客戶端-伺服器之實施上傳送。感應器實施伺服器。嵌入式的HTTP伺服器直接提供使用者介面給組態與診斷用之儀器。應用XML之命令與回應的指令表允許資料收集客戶端自動發現網路上的這些感應器、擷取供收集之測量(亦即變量)之完整的表單、界定專門應用收集計畫以及擷取經緩衝的資料取樣。Several sensors and material transfer products are implemented in various development stages to perform sampling buffering. They use the data model of the application XML, which is passed on the implementation of the HTTP stateless client-server. The sensor implements a server. The embedded HTTP server directly provides the user interface to the instrument for configuration and diagnostics. The XML command and response command form allows the data collection client to automatically discover these sensors on the network, retrieve complete forms of measurements (ie, variables) for collection, define specialized application collection plans, and capture buffered Data sampling.
這些感應器的一些使用不同的介面供控制用,其於某些情況中使用個別的實際DeviceNet連結,同時其他與MODBUS/TCP分享相同的乙太網路埠。已證實HTTP資料收集與使用者互動不會導致統計上可重複的性能或定時之失敗。Some of these sensors use different interfaces for control, in some cases using individual actual DeviceNet links, while others share the same Ethernet network with MODBUS/TCP. It has been confirmed that HTTP data collection and user interaction does not result in statistically repeatable performance or timing failures.
為了實施程序I/O控制器的兩種特性-硬性即時異動導向網路I/O控制以及收集之資料的時間彈性的發表-這些感應器之特徵在於先進的即時軟體基礎。根據此基礎,這兩種應用皆在對資源之利用嚴格掌控以及資源分配之有意識的優先權排序的方式下建立,使得在允許的效能參數內之資料收集活動不會降低MODBUS之控制命令效能。並且,實施收集到之資料的發表,使得在預期的MODBUS活動量下,有足夠的剩餘頻寬可用來發表收集到之資料而不會損失品質或量。In order to implement the two features of the program I/O controller - hard real-time traffic-oriented network I / O control and the publication of the time elasticity of the collected data - these sensors are characterized by an advanced real-time software foundation. Based on this foundation, both applications are established in a manner that strictly controls the use of resources and prioritizes the allocation of resources, so that the data collection activities within the allowed performance parameters do not reduce the control command performance of MODBUS. . Moreover, the publication of the collected data enables sufficient residual bandwidth to be used to publish the collected data without loss of quality or quantity under the expected MODBUS activity.
除了優先權排序之外,變量取樣之總數以及收集的頻率係有限的,使得當資料收集活動與MODBUS結合時不會過度使用可得之資源,導致收集資料的遺失或導致不可預期之系統行為。In addition to prioritization, the total number of variable samples and the frequency of collection are limited, so that when data collection activities are combined with MODBUS, the available resources are not overused, resulting in the loss of collected data or unpredictable system behavior.
針對控制之目的,程序I/O控制器允許藉由連接器接腳數對其之I/O之低層級的存取,以及值係由簡單的二進位形式代表。預期工具主機執行連接器接腳數對變量之其本身的映照,而CONTROLweb不會執行任何轉譯以保留原始值並且將效能最大化。For control purposes, the program I/O controller allows access to the lower level of its I/O by the number of connector pins, and the value is represented by a simple binary form. It is expected that the tool host will perform the mapping of the number of connector pins to the variables themselves, and CONTROLweb will not perform any translations to preserve the original values and maximize performance.
另一方面,針對分析,資料元件應具有邏輯意義-變量名字應代表實際參數,以及可按比例調整類比值之大小並且偏移至測量的實體單位。應針對感應器校準補償值,使得感應器收集之來自類似處理室的值可被比較。On the other hand, for analysis, the data element should have a logical meaning - the variable name should represent the actual parameter, and the size of the analog value can be scaled and offset to the measured physical unit. The compensation values should be calibrated for the sensor so that the values collected by the sensor from similar processing chambers can be compared.
欲增加連接器接腳之符號式代表,這些程序I/O控制器實施變量映照表,其為可組態並儲存於非揮發性記憶體中,其分配邏輯名字以及數字調整到關注之I/O信號。注意到針對某些程序I/O控制器,I/O接腳數超過映照表中之符號變量之數量。若僅可存取映照之I/O接腳,映照表亦辨別關注之I/O信號,以供資料收集與分析。To increase the symbolic representation of the connector pins, these program I/O controllers implement a variable mapping table that is configurable and stored in non-volatile memory, with assigned logical names and numbers adjusted to the I/I of interest. O signal. Note that for some program I/O controllers, the number of I/O pins exceeds the number of symbol variables in the map. If only the mapped I/O pins are accessible, the mapping table also identifies the I/O signals of interest for data collection and analysis.
程序I/O控制器係針對附接至I/O接腳之感應器與控制而組態。辨別並命名關注之變量,且必要時施加數字的按比例調整之偏移,使報告的值更有意義,較佳以測量之標準單位。由於以符號命名收集的資料,仍可識別其之意義即使當資料拿離線或離場(off-site),其中僅能取得給定工具設定之少量的細節。收集到的值可長時間保留他們的意義,即使當時實際的工具作升級或更新。The program I/O controller is configured for sensors and controls attached to the I/O pins. Identify and name the variables of interest, and if necessary, apply a scaled offset of the numbers to make the reported values more meaningful, preferably in standard units of measurement. Since the data collected by symbolic naming still recognizes its meaning even when the material is taken offline or off-site, only a small amount of detail can be obtained for a given tool. The collected values can retain their meaning for a long time, even if the actual tools are upgraded or updated at the time.
當組態安裝於類似工具上之程序I/O控制器時,能夠藉由下載組態到文字檔並上載文字檔到其他的程序I/O控制器而簡化組態。這些組態檔可在離線時使用標準PC軟體工具作編輯,使得可由處理專家在離場狀態準備組態。When configuring a program I/O controller installed on a similar tool, you can simplify the configuration by downloading the configuration to a text file and uploading the text file to other program I/O controllers. These configuration files can be edited offline using standard PC software tools so that the configuration can be prepared by the processing expert in the off-going state.
卡片機架控制器之架構最初的目標係塞入MODBUS於乙太網路上可能之最快的回應時間,並改善控制之可重複的準確度超越諸如DeviceNet之較慢的商業匯流排可達到之準確度。於許多情況中已學習到,處理準確度之定時之瓶頸在於工具主機或主控制器本身以及用來植入工具主機之操作系統。某些現有的主控制器在50毫秒之容限內操作,其他在100毫秒的容限內。The initial goal of the card rack controller architecture is to insert the fastest possible response time on MODBUS on the Ethernet and improve the repeatability of the control beyond the slower commercial bus such as DeviceNet. degree. It has been learned in many cases that the bottleneck of the timing of processing accuracy lies in the tool host or the main controller itself and the operating system used to implant the tool host. Some existing main controllers operate within a tolerance of 50 milliseconds, others are within a tolerance of 100 milliseconds.
實際上,這些控制系統頗為複雜且經常由龐大的設計團隊長時間開發出來。使用的開發工具僅能在一般目的之計算平台上取得。這些舊有軟體的複雜度係導致其之不準確性之主因。In fact, these control systems are complex and often developed over time by a large design team. The development tools used can only be obtained on a general purpose computing platform. The complexity of these old softwares is the main cause of their inaccuracy.
欲改善處理的準確度而不給予每一個處理工具一個專屬的工作站,這些程序I/O控制器實施有限但功能強大之編程環境以控制遠端I/O。這些編程環境並非意圖取代傳統的處理控制軟體系統,而係藉由委派處理控制邏輯那些具有關鍵定時需求的部分給以即時為基礎的程序I/O控制單元而輔助它們,即時為基礎的程序I/O,控制單元維持在工具主機或主控制器的控制之下。To improve the accuracy of the processing without giving each processing tool a dedicated workstation, these program I/O controllers implement a limited but powerful programming environment to control remote I/O. These programming environments are not intended to replace traditional processing control software systems, but rather assist them by delegating processing control logic to those parts with critical timing requirements to instant-based program I/O control units, instant-based programs I /O, the control unit is maintained under the control of the tool host or the main controller.
軟體或操作員辨別出小但關鍵的處理步驟,並將需要的監視與反饋轉譯成可在資源有限且確定性之操作環境內執行之小程式,該操作環境具有定時準確性,諸如優於5毫秒、優於3毫秒或優於1毫秒之準確性。The software or operator identifies small but critical processing steps and translates the required monitoring and feedback into small programs that can be executed within a resource-constrained and deterministic operating environment with timing accuracy, such as better than 5 The accuracy of milliseconds, better than 3 milliseconds or better than 1 millisecond.
程序I/O控制器區域編程支援ECMA標準Javascript語言之子集,其由物件導向事件驅動之解譯的程式語言。以完全文字下載一程式到I/O單元中,之後可使用特別分配之程式狀態暫存器喚起該程式,其可由MODBUS設定並監視。此程式狀態暫存器亦用於工具主機以監視該程式之執行與完成。適合實施此環境的一處理器家族為精簡指令集計算(RISC)微處理器之Motorola ColdFire家族。此家族中之微處理器之指定的MCF5282。由於這些處理器係衍生自可敬的68000處理器系列,有許多開發工具以及各種即時操作系統。此處理器家族之區塊圖與細節可從www.freescale.com取得。Program I/O Controller Area Programming supports a subset of the ECMA standard Javascript language, which is an object-oriented event-driven interpreted programming language. Download a program to the I/O unit in full text, and then use the specially assigned program status register to evoke the program, which can be set and monitored by MODBUS. This program status register is also used by the tool host to monitor the execution and completion of the program. One processor family suitable for implementing this environment is the Motorola ColdFire family of Reduced Instruction Set Computing (RISC) microprocessors. The designated MCF5282 of the microprocessor in this family. Because these processors are derived from the respectable 68000 processor family, there are many development tools and various real-time operating systems. Block diagrams and details of this processor family are available at www.freescale.com.
在任何給定時間應執行有限數量之指令碼程式,並且這些程式應有尺寸限制。引進尺寸限制並且從此實施移除某些語言工具或特徵以減低使用者編寫之指令碼會超過可用的計算資源之危險,並導致控制子系統之未界定的行為。雖然,仍可能經由編程錯誤而使程序I/O控制器之分配到比指令碼編譯器優先權更低之某些地方部分混亂。有危險的特徵包含網路使用者介面、資料收集以及運作其他並行之指令碼的相鄰的指令碼編譯器。應此建議將這些使用者指令碼程式維持短且簡單,避免緊湊的迴路並在可能之處使用事件通知。A limited number of scripts should be executed at any given time, and these programs should be limited in size. Introducing size limitations and removing certain language tools or features from this implementation to reduce the risk that user-written scripts will exceed available computing resources and result in undefined behavior of the control subsystem. Although it is still possible to programmatically assign a program I/O controller to a portion of the lower order priority than the instruction code compiler via programming errors. Hazardous features include a network user interface, data collection, and adjacent script compilers that run other parallel scripts. It is recommended to keep these user scripts short and simple, avoiding tight loops and using event notification where possible.
特別的網頁將顯示最近的指令碼執行歷史。此網頁提供比在MODBUS上可得之程式狀態暫存器更多之細節。A special web page will display the most recent script execution history. This page provides more details than the program status register available on MODBUS.
於指令碼編譯器內建立一階級制度以代表程序I/O控制器上之區域I/O信號,使得指令碼可對輸出行使控制並監視輸入。並且,離散之輸入可產生事件,其可被分配到特定的指令碼功能,並且當這些輸入改變值或狀態時非同步地喚起這些功能。A class system is established within the instruction code compiler to represent the regional I/O signals on the program I/O controller so that the instruction code can exercise control over the output and monitor the input. Also, discrete inputs can generate events that can be assigned to particular instruction code functions and that are asynchronously evoked when these inputs change values or states.
指令碼編譯器分配有可透過MODBUS讀取與寫入之特別16位元暫存器,以控制並監視對應指令碼編譯器之狀態。主或中央控制器可寫入此暫存器以開始或終止一指令碼的執行,或讀取此暫存器以監視編譯器的狀態,亦即,若指令碼正在運作,或已正常結束,或非正常結束,於後者情況中此暫存器將會包含指令碼退離碼。The instruction code compiler is assigned a special 16-bit scratchpad that can be read and written via MODBUS to control and monitor the state of the corresponding instruction code compiler. The master or central controller can write to the scratchpad to start or terminate the execution of an instruction code, or read the scratchpad to monitor the state of the compiler, that is, if the instruction code is operating, or has ended normally, Or the abnormal end, in the latter case, the register will contain the instruction code exit code.
可進一步通用上述的程式控制暫存器,並且針對每一個指令碼編譯器可預先界定有限數量之額外的暫存器。這些額外的暫存器,如同程式狀態暫存器,可透過MODBUS進行讀取與寫入之存取,並可由在適當的編譯器上運作之指令碼存取。這些暫存器可用於涉及運行在CONTROLweb上之指令碼與主控制器之間的通訊之各種目的。The program control registers described above can be further generalized, and a limited number of additional registers can be predefined for each instruction code compiler. These additional registers, like the program status register, can be accessed by MODBUS for read and write access and can be accessed by instruction code running on the appropriate compiler. These registers can be used for various purposes involving communication between the script running on CONTROLweb and the host controller.
除了對區域I/O信號之完整不受阻礙之存取之外,指令碼編譯器添加有階級制度,其可經設定而代表與其他程序I/O控制器實體連接之I/O信號,但在經關聯後表現幾乎像區域I/O信號,具有相同的基礎存取方法與屬性。位在其他程序I/O控制器上之I/O信號(遠端I/O操作)通常較慢回應並無法作為非同步事件的來源。In addition to complete unobstructed access to regional I/O signals, the script compiler adds a class system that can be set to represent I/O signals that are connected to other program I/O controller entities, but After association, it behaves almost like a regional I/O signal, with the same basic access methods and attributes. I/O signals (remote I/O operations) located on other program I/O controllers are usually slower to respond and cannot be used as a source of asynchronous events.
藉由添加MODBUS客戶端(主)實施到指令碼編譯器而實現此在其他程序I/O控制器上之遠端I/O,該實施提供對連接到相同網路之其他程序I/O控制器上之I/O信號的存取。This remote I/O on other program I/O controllers is implemented by adding a MODBUS client (primary) implementation to the instruction code compiler, which provides additional program I/O control to the same network. Access to I/O signals on the device.
考量到上述限制,應謹慎使用此便利的設備。最佳用途之一為將在不同程序I/O控制器上並行之指令碼同步化,其結合以執行單一處理序列,其中I/O數量以及/或處理實體佈局為將所有有關之I/O信號連接至相同的程序I/O控制器係不可能的。Taking into account the above restrictions, this convenient device should be used with caution. One of the best uses is to synchronize parallel scripts on different program I/O controllers, which combine to perform a single processing sequence where the number of I/Os and/or processing entities are laid out for all relevant I/O. It is not possible to connect signals to the same program I/O controller.
一製造處理細節係通常稱為“配方”,其事實上為一主導包含化學或區段處理之序列、時間以及臨限值之細節的程式。於一些先前的處理器中,藉由工廠控制主機下載配方到室處理控制電腦(或選擇先前儲存之配方)。接著由主控制器對一批次中之每一個單一的晶圓執行此配方。主控制器的目標為僅可能準確地對一給定批次中之每一個晶圓,或任何其他批次,重複此配方,使得每次出來的處理結果(如沉積層之厚度)皆相同。但如同先前所述,根據具有大且複雜之軟體系統之一般目的的計算平台之目前的處理控制系統在定時之準確度上有其可辨別之限制,並且對於輸入改變之快速回應並且夠快反應的能力有限。這些限制最終抑制配方可重複之準確度,且為嚴重限制目前處理設配良率與產量之因素。因此,故意將某些處理速度減慢以將之及時停止。A manufacturing process detail is often referred to as a "recipe," which is in fact a program that dominates the sequence, time, and threshold details of the chemical or segmentation process. In some previous processors, the factory control host downloaded the recipe to the room to process the control computer (or select a previously stored recipe). This recipe is then executed by the host controller for each of a single wafer in a batch. The primary controller's goal is to repeat this recipe only for each wafer in a given batch, or any other batch, so that the results of each process (such as the thickness of the deposited layer) are the same. However, as previously stated, current processing control systems based on computing platforms having the general purpose of large and complex software systems have discretionary limitations in timing accuracy, and respond quickly to input changes and react quickly enough. Limited ability. These limits ultimately inhibit the repeatability of the formulation and are a significant constraint on the current processing yield and yield factors. Therefore, some processing speeds are deliberately slowed down to stop them in time.
以這些改良之程序I/O控制器對此問題提出之解決方案為有效地委派配方那些視為關鍵的部份到以即時為基礎之遠端I/O裝置。The solution to this problem with these improved program I/O controllers is to effectively delegate recipes that are considered critical to instant I/O devices based on instant.
主處理控制器,取代迭代地一次處理一個配方指令(為目前的標準作業),會先整理一配方,並界定需要定時以及回應準確性而主控制器能力無法及之那些部分。這些關鍵配方步驟將編譯成為指令碼(Script),並載入適當的程序I/O控制器以供後續執行。接著,程序I/O控制器將迭代地執行該配方:執行那些未被用自己的資源標示為“關鍵”的部分,並當到達“關鍵步驟”時,起始指令碼以執行遠端I/O單元之一,並且經由一或更多虛擬I/O變量監視指令碼之執行,這些變量由指令碼用來指示其之狀態。The main processing controller, instead of iteratively processing one recipe instruction at a time (for current standard jobs), will first organize a recipe and define those parts that require timing and response accuracy that are beyond the capabilities of the main controller. These key recipe steps are compiled into scripts and loaded into the appropriate program I/O controller for subsequent execution. Next, the program I/O controller will iteratively execute the recipe: execute those parts that are not marked as "critical" by their own resources, and when they reach the "critical step", start the instruction code to execute the remote I/ One of the O units, and the execution of the instruction code is monitored via one or more virtual I/O variables that are used by the instruction code to indicate its status.
在編譯期間,主控制器將所有對邏輯值與狀態的參照替換成實體或映照I/O信號參照,不然則將配方步驟轉換成指令碼程式。達成此之一種方法係使用指令碼模板,其中預先存在之指令碼係寫給特定之配方步驟種類,並且主控制器將接著需要置入實際的I/O參考以及具體的時間與臨限值到預定之指令碼位置。During compilation, the host controller replaces all references to logic values and states to physical or mapped I/O signal references, otherwise the recipe steps are converted to instruction code programs. One way to achieve this is to use a script template in which pre-existing instruction codes are written to a particular recipe step type, and the main controller will then need to place the actual I/O reference and the specific time and threshold to The predetermined command code location.
本發明可以方法或調適成實施該方法之裝置實施。可從工具主機、僕或程序I/O控制器以及操作協定或系統的態樣來看待相同之方法。本發明可為製品,如紀錄有邏輯之媒體以執行方法。The invention can be practiced or adapted to implement the apparatus. The same approach can be seen from the tool host, servant or program I/O controller, and the mode of operation agreement or system. The invention can be an article of manufacture, such as recording a logical medium to perform the method.
一實施例為對半導體製造程序之準確定時之控制的方法。此方法使用工具主機以及程序I/O控制器。該程序I/O控制器包含針對處理室監視以及控制之電性介面,並透過電性介面支援準確定時之輸入與輸出。工具主機包含符號式代表電性介面之編程環境。此方法包含,在工具主機,準備包含使該程序I/O控制器經由該電性介面取樣輸入並控制輸出之指令的控制程式。從該工具主機載入該控制程式至該程序I/O控制器。在該程序I/O控制器,當接收到喚起該控制程式之命令,運作該控制程式以經由該電性介面產生取樣輸入以及控制輸出的統計上精確的定時。當在目標時間以及實際時間之間的變異分布之百分之99.99的範圍中的變異小於或等於五毫秒時,該取樣輸入以及該控制輸出被視為具有統計上精確之定時。於一些範例中,目標時間可為儘可能的快並且一命令之執行可在小於或等於五毫秒。於其他範例中,目標時間可為當輸入達到特定程度時。One embodiment is a method of controlling the timing of semiconductor fabrication processes. This method uses a tool host and a program I/O controller. The program I/O controller includes an electrical interface for processing and monitoring of the processing room, and supports accurate timing input and output through the electrical interface. The tool host contains a symbolic representation of the programming environment of the electrical interface. The method includes, at a tool host, preparing a control program including instructions for causing the program I/O controller to sample input and control output via the electrical interface. The control program is loaded from the tool host to the program I/O controller. At the program I/O controller, upon receiving a command to evoke the control program, the control program is operated to generate a sample input and a statistically accurate timing of the control output via the electrical interface. The sample input and the control output are considered to be statistically accurate when the variation in the range of 99.99 percent of the variation distribution between the target time and the actual time is less than or equal to five milliseconds. In some examples, the target time may be as fast as possible and the execution of a command may be less than or equal to five milliseconds. In other examples, the target time can be when the input reaches a certain level.
此方法之一態樣為可在諸如Javascript之解譯式程式語言中準備控制程式。使用解譯式程式語言促進控制程式之載入並不擾亂執行命令或其他程式之優先順序。另一態樣,其可與第一結合,為控制程式之載入以載入處理優先權進行,其保留程序I/O控制器之處理命令的優先權,無論從其他程序I/O控制器或工具主機接收到。又一態樣涉及載入控制程式而不將程序I/O控制器從透過電性介面之至少一些取樣輸入以及控制輸出離線。One aspect of this approach is to prepare the control program in an interpreted programming language such as Javascript. Using an interpreted program language to facilitate loading of control programs does not disturb the order of execution of commands or other programs. In another aspect, it can be combined with the first to load the control program to load processing priority, which retains the priority of the processing command of the program I/O controller, regardless of other program I/O controllers. Or the tool host received it. Yet another aspect involves loading the control program without taking the program I/O controller offline from at least some of the sample inputs and control outputs through the electrical interface.
於至少一實施例中,以小於或等於3(三)毫秒之目標時間與實際時間之間的變異分布之百分之99.99的範圍執行取樣之輸入以及該控制之輸出。所述的單石控制器實施例可提供取樣輸入以及控制輸出之精確的定時,使得目標時間以及實際時間之間的變異分布之百分之99.99的範圍係小於或等於一毫秒In at least one embodiment, the input of the sample and the output of the control are performed in a range of 99.99 percent of the distribution of variations between the target time and the actual time of less than or equal to 3 (three) milliseconds. The single stone controller embodiment can provide accurate timing of the sampling input and control output such that the range of 99.99 percent of the distribution of variation between the target time and the actual time is less than or equal to one millisecond.
於操作中,此方法包含程序I/O控制器透過電性介面設定工具主機對應取樣的輸入與改變之輸出的虛擬數位信號。程序I/O控制器發送這些虛擬數位信號而不擾亂取樣與控制之時間的統計上之精確性。虛擬數位信號可包含時間戳。工具主機,回應虛擬數位信號,可發送一命令至程序I/O控制器以令程序I/O控制器停止運作控制程式。In operation, the method includes the program I/O controller setting the virtual digital signal of the corresponding input and the changed output of the tool host through the electrical interface. The program I/O controller sends these virtual digital signals without disturbing the statistical accuracy of the sampling and control time. The virtual digit signal can include a timestamp. The tool host, in response to the virtual digit signal, can send a command to the program I/O controller to cause the program I/O controller to stop operating the control program.
程序I/O控制器可,回應控制程式,以統計上精確的時間發送控制命令至另一程序I/O控制器。工具主機可組態成接受一配方中之控制步驟的定時容限並判斷是否委派那些控制步驟之時間至程序I/O控制器。當判斷欲委派控制步驟的時間時,可準備一控制程式並載入到程序I/O控制器。The program I/O controller can, in response to the control program, send control commands to another program I/O controller at a statistically accurate time. The tool master can be configured to accept the timing tolerances of the control steps in a recipe and determine whether to delegate those control steps to the program I/O controller. When judging the time to delegate the control step, a control program can be prepared and loaded into the program I/O controller.
另一實施例為使用與中央控制器配合之程序I/O控制器在重複間隔來控制運作於處理室中之程序的方法。此方法,包含,在監視並控制操作的處理室之一或更多態樣之該程序I/O控制器,自中央控制器接收控制命令並在接收後五毫秒或更少的第一統計上可重複的短間隔內處理該控制命令。當在時間分布的百分之99.99的範圍中的時間小於或等於該短間隔時,短間隔係視為統計上可重複的。程序I/O控制器,在不失去控制命令的處理中之統計上的可重複性的情況下,亦取樣耦合至處理室之一或更多感應器並於取樣排程表之統計上可重複的容限內提供該取樣。當在目標時間以及實際時間之間的變異分布之百分之99.99的範圍中的變異小於或等於該容限時,該容限係視為統計上可重複的。取決於硬體實施例,第一統計上可重複的短間隔可為3 ms或更少或1 ms或更少。同樣取決於硬體實施例,統計上可重複之容限可為五毫秒、三毫秒或一毫秒或更少。Another embodiment is a method of controlling a program operating in a processing chamber at repeated intervals using a program I/O controller in conjunction with a central controller. The method includes, in monitoring and controlling one or more aspects of an operating chamber of the program, the program I/O controller receives a control command from the central controller and receives a first statistic of five milliseconds or less after receiving The control command is processed in a repeatable short interval. When the time in the range of 99.99 percent of the time distribution is less than or equal to the short interval, the short interval is considered to be statistically repeatable. The program I/O controller, also sampling one or more sensors coupled to the processing chamber and statistically repeatable in the sampling schedule, without statistically repeatable in the processing of the control command This sampling is provided within the tolerances. The tolerance is considered to be statistically repeatable when the variation in the range of 99.99 percent of the distribution of variation between the target time and the actual time is less than or equal to the tolerance. Depending on the hardware embodiment, the first statistically repeatable short interval may be 3 ms or less or 1 ms or less. Also depending on the hardware embodiment, the statistically repeatable tolerance can be five milliseconds, three milliseconds, or one millisecond or less.
實施上述方法,緩衝取樣可包含緩衝對應取樣之時間戳。程序I/O控制器可分布該經緩衝之取樣的至少一些而不失去該控制命令的處理之統計上的可重複性或失去該取樣之統計上的可重複性。取而代之,此方法進一步包含,在不失去該控制命令的處理之可重複性或失去該取樣之統計上的可重複性的情況下,從該程序I/O控制器之記憶體動態執行可載入之指令。這些指令可包含回應該取樣之選定的結果,該回應在第二統計上可重複的短間隔內發生,該第二統計上可重複的短間隔小於將取樣報告給中央控制器並接收回回應的命令所需之統計上可重複的反饋間隔。取而代之,該第二統計上可重複的短間隔可小於該第一統計上可重複的短間隔之一半長。Implementing the above method, buffer sampling may include buffering the timestamp of the corresponding sample. The program I/O controller may distribute at least some of the buffered samples without losing statistical reproducibility of the processing of the control commands or losing statistical reproducibility of the samples. Instead, the method further includes dynamically loading the load from the memory of the program I/O controller without losing the repeatability of the processing of the control command or losing the statistical reproducibility of the sample Instructions. The instructions may include a result of the selection that should be sampled, the response occurring within a second statistically repeatable short interval that is less than reporting the sample to the central controller and receiving the response. The statistically repeatable feedback interval required for the command. Alternatively, the second statistically repeatable short interval may be less than one-and-a-half the length of the first statistically repeatable short interval.
此方法進一步包含,在不將程序I/O控制器自操作處理室之監視與控制態樣離線的情況下,接收、辨識以及動態載入指令到程序I/O控制器之記憶體中。替代地或結合地,針對它們該方法可包含程序I/O控制器分布至少一些經緩衝的取樣而不失去控制命令之接收與處理之統計上的可重複性或失去於載入指令中之動態執行之統計上的可重複性。The method further includes receiving, identifying, and dynamically loading the instructions into the memory of the program I/O controller without taking the program I/O controller offline from the monitoring and control aspects of the operating processing chamber. Alternatively or in combination, the method for the method may include the program I/O controller distributing at least some of the buffered samples without losing statistical reproducibility of the receipt and processing of the control commands or loss of dynamics in the load instruction Statistical reproducibility of implementation.
此方法的另一態樣可為,不失去控制命令之接收與處理之統計上的可重複性或失去於載入指令中之動態執行之統計上的可重複性,程序I/O控制器從記憶體動態執行載入之指令。詳言之,這些指令可起始包含操作處理室之至少一態樣的閉路控制,該起始在小於反饋間隔或小於該第一統計上可重複的短間隔之一半長之第二統計上可重複的短間隔內發生。於一些範例中,第二短間隔可小於第一統計上可重複的短間隔之五分之一長。此方法可包含,程序I/O控制器接收、辨識以及動態載入指令到程序I/O控制器之記憶體中而不將程序I/O控制器自操作處理室之監視與控制態樣離線。程序I/O控制器分布至少一些經緩衝的取樣而不失去統計上的可重複性。Another aspect of the method may be that the statistical repeatability of the reception and processing of the control command is lost or the statistical reproducibility of the dynamic execution lost in the load instruction, the program I/O controller is The memory dynamically executes the load instruction. In particular, the instructions may initiate a closed loop control comprising at least one aspect of the operational processing chamber, the initiation being second statistically less than a feedback interval or less than one of the first statistically repeatable short interval Repeated within a short interval. In some examples, the second short interval may be less than one fifth of the first statistically repeatable short interval. The method can include the program I/O controller receiving, identifying, and dynamically loading instructions into the memory of the program I/O controller without taking the program I/O controller from the monitoring and control mode of the operating processing room offline. . The program I/O controller distributes at least some of the buffered samples without losing statistical reproducibility.
作為一裝置,程序I/O控制器可包含埠,其調適成與耦合至埠之中央處理器、記憶體以及邏輯與資源通訊,並且記憶體調適成執行方法以及上述方法之任何態樣。As a device, the program I/O controller can include a device that is adapted to communicate with a central processing unit, memory, and logic and resources coupled to the device, and the memory is adapted to perform the method and any aspect of the method described above.
雖以參照上述較佳實施例與範例揭露本發明,可了解到這些範例意圖為例示性而非限制性者。The present invention has been described with reference to the preferred embodiments and examples, which are intended to be illustrative and not restrictive.
111...半導體製造廠網路111. . . Semiconductor manufacturing network
112...網路112. . . network
113...氣體箱壓力變換器113. . . Gas box pressure transducer
114...質量流控制器(MFC)114. . . Mass flow controller (MFC)
115...反應性氣體產生器115. . . Reactive gas generator
116...晶圓監視器116. . . Wafer monitor
117...氣體成分監視器117. . . Gas composition monitor
125...處理室125. . . Processing room
126...窗戶126. . . window
127...處理監視器127. . . Processing monitor
133...臭氧產生器133. . . Ozone generator
134...材料遞送系統134. . . Material delivery system
136...工作件136. . . Work piece
137...壓力變換器137. . . Pressure transducer
142...數位控制器142. . . Digital controller
145...冷卻子系統145. . . Cooling subsystem
146...洩漏偵測器146. . . Leak detector
147...控制閥147. . . Control valve
148...壓力控制器148. . . pressure controller
152...室控制器152. . . Room controller
153...微瓦電源153. . . Microwatt power supply
154...RF電源154. . . RF power supply
155...DC電源155. . . DC power supply
156...子系統156. . . Subsystem
157...真空計157. . . Vacuum gauge
158...排出監視器158. . . Discharge monitor
162...連接點162. . . Junction
210...曲線210. . . curve
211,222,224...截斷線211,222,224. . . Chop line
220...曲線220. . . curve
223...目標時間223. . . Target time
311...工具主機311. . . Tool host
312、316...SEC/GEM介面埠312, 316. . . SEC/GEM interface埠
315...SECS多工器315. . . SECS multiplexer
317...SEC/GEM工具介面317. . . SEC/GEM tool interface
321...報告321. . . report
322...網路322. . . network
325...資料收集與發表資源325. . . Data collection and publication resources
327...感應器327. . . sensor
331...貯存331. . . Storage
332、336...網路介面埠332, 336. . . Network interface埠
334...半導體廠側介面334. . . Semiconductor factory side interface
335...工具側介面335. . . Tool side interface
337...儀器337. . . instrument
401...工具401. . . tool
402...感應器402. . . sensor
402...A-B感應器402. . . A-B sensor
403...通訊協調器403. . . Communication coordinator
407、409...SECS協定407, 409. . . SECS Agreement
408...連結408. . . link
509...工具側網路509. . . Tool side network
513...管理的交換器513. . . Managed switch
515...網路附接儲存器(NAS)515. . . Network Attached Storage (NAS)
519...半導體廠側網路519. . . Semiconductor factory side network
523...分析軟體523. . . Analysis software
525...延伸網路525. . . Extended network
527...資料庫527. . . database
611...連鎖卡611. . . Chain card
612...數位I/O612. . . Digital I/O
613...類比I/O613. . . Analog I/O
621...SCAN匯流排621. . . SCAN bus
631...介面控制器631. . . Interface controller
642...前端642. . . front end
643...後端643. . . rear end
711...即時控制711. . . Instant control
712...資料記錄器712. . . Data logger
713...診斷器713. . . Diagnostic
723...資料收集723. . . data collection
724...使用者介面724. . . user interface
731...I/O驅動器731. . . I/O driver
732...閘道732. . . Gateway
733...MODBUS TCP驅動器733. . . MODBUS TCP driver
734...HTTP伺服器734. . . HTTP server
741...I/O通道741. . . I/O channel
742...UART742. . . UART
743...TCP/IP743. . . TCP/IP
753...乙太網路753. . . Ethernet
810...工具主機810. . . Tool host
811...控制程式811. . . Control program
812...MODBUS/TCP主應用程式介面812. . . MODBUS/TCP main application interface
813...網路瀏覽器813. . . Web browser
815...乙太網路815. . . Ethernet
820、830...程序I/O控制器820, 830. . . Program I/O controller
821...MODBUS/TCP介面821. . . MODBUS/TCP interface
822...網路伺服器822. . . Web server
823...Javascript常規823. . . Javascript convention
824...資料收集824. . . data collection
825...I/O介面接腳825. . . I/O interface pin
第1圖描述其中本發明之態樣特別有用之環境。Figure 1 depicts an environment in which the aspects of the invention are particularly useful.
第2A-2B圖以分布圖描述效能與變異統計。Figure 2A-2B depicts the performance and variation statistics in a distribution map.
第3圖為使用單一通訊通道與工具、感應器以及工具主機通訊之程序I/O控制器的方塊圖。第4圖描述多種通訊通道之使用。Figure 3 is a block diagram of a program I/O controller that communicates with tools, sensors, and tool hosts using a single communication channel. Figure 4 depicts the use of various communication channels.
第5圖為可用來建立程序I/O控制器之軟體與硬體零件的方塊圖。Figure 5 is a block diagram of the software and hardware components that can be used to build a program I/O controller.
第6圖為程序I/O控制器實際架構之高度方塊圖。Figure 6 is a block diagram of the height of the actual architecture of the program I/O controller.
第7圖為具有MODBUS介面之I/O與COM架構的方塊圖。Figure 7 is a block diagram of the I/O and COM architecture with MODBUS interface.
第8圖為將程序I/O控制器關聯到工具主機之高度方塊圖。Figure 8 is a block diagram showing the height of the program I/O controller associated with the tool host.
711...即時控制711. . . Instant control
712...資料記錄器712. . . Data logger
713...診斷器713. . . Diagnostic
723...資料收集723. . . data collection
724...使用者介面724. . . user interface
731...I/O驅動器731. . . I/O driver
732...閘道732. . . Gateway
733...MODBUS TCP驅動器733. . . MODBUS TCP driver
734...HTTP伺服器734. . . HTTP server
741...I/O通道741. . . I/O channel
742...UART742. . . UART
743...TCP/IP743. . . TCP/IP
753...乙太網路753. . . Ethernet
Claims (31)
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US67677005P | 2005-05-02 | 2005-05-02 |
Publications (2)
Publication Number | Publication Date |
---|---|
TW200709009A TW200709009A (en) | 2007-03-01 |
TWI397830B true TWI397830B (en) | 2013-06-01 |
Family
ID=49030111
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
TW95115310A TWI397830B (en) | 2005-05-02 | 2006-04-28 | Versatile semiconductor manufacturing controller with statistically repeatable response times |
Country Status (1)
Country | Link |
---|---|
TW (1) | TWI397830B (en) |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
TW418454B (en) * | 1999-02-09 | 2001-01-11 | Samsung Electronics Co Ltd | Semiconductor device manufacturing apparatus having controller detecting function of filter |
TW533457B (en) * | 2000-06-16 | 2003-05-21 | Tokyo Electron Ltd | Semiconductor manufacturing apparatus detecting a state of connection between controllers |
US6759334B2 (en) * | 1999-09-01 | 2004-07-06 | Canon Kabushiki Kaisha | Semiconductor manufacturing apparatus and method |
-
2006
- 2006-04-28 TW TW95115310A patent/TWI397830B/en not_active IP Right Cessation
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
TW418454B (en) * | 1999-02-09 | 2001-01-11 | Samsung Electronics Co Ltd | Semiconductor device manufacturing apparatus having controller detecting function of filter |
US6254767B1 (en) * | 1999-02-09 | 2001-07-03 | Samsung Electronics Co., Ltd. | Semiconductor device manufacturing apparatus having controller detecting function of filter |
US6759334B2 (en) * | 1999-09-01 | 2004-07-06 | Canon Kabushiki Kaisha | Semiconductor manufacturing apparatus and method |
TW533457B (en) * | 2000-06-16 | 2003-05-21 | Tokyo Electron Ltd | Semiconductor manufacturing apparatus detecting a state of connection between controllers |
Also Published As
Publication number | Publication date |
---|---|
TW200709009A (en) | 2007-03-01 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
JP4928541B2 (en) | Versatile semiconductor manufacturing controller with statistically repeatable response time | |
CN100375086C (en) | Method and device for dynamic sensor configuration and runtime execution | |
JP6904639B2 (en) | Background collection of diagnostic data from field instrumentation | |
US7693687B2 (en) | Controller and method to mediate data collection from smart sensors for fab applications | |
US6952656B1 (en) | Wafer fabrication data acquisition and management systems | |
US8812261B2 (en) | Method and apparatus to automatically create virtual sensors with templates | |
JP4685446B2 (en) | How to process data based on the data context | |
KR100873114B1 (en) | Method and apparatus for automatic sensor installation | |
US7756657B2 (en) | System for storing and presenting sensor and spectrum data for batch processes | |
EP1508159A2 (en) | Method and apparatus for monitoring tool performance | |
EP2490088A2 (en) | Distributed proportional/integral/derivative tuning | |
EP1373997B1 (en) | Method and apparatus for controlling a tool using a baseline control script | |
WO2001052320A2 (en) | A method for requesting trace data reports from fault detection controlled semiconductor fabrication processes | |
EP1415126B1 (en) | Method and apparatus for integrating multiple process controllers | |
TWI397830B (en) | Versatile semiconductor manufacturing controller with statistically repeatable response times | |
KR101071077B1 (en) | Architecture for general purpose programmable semiconductor processing system and methods therefor | |
Miletic et al. | Automated performance tracking of sensors and control loops | |
KR100970684B1 (en) | Method for dynamic sensor configuration and runtime execution |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
MM4A | Annulment or lapse of patent due to non-payment of fees |