TW200404333A - Multi-variable analysis model forming method of processing apparatus, multi-variable analysis method for processing apparatus, control apparatus of processing apparatus, and control system of processing apparatus - Google Patents
Multi-variable analysis model forming method of processing apparatus, multi-variable analysis method for processing apparatus, control apparatus of processing apparatus, and control system of processing apparatus Download PDFInfo
- Publication number
- TW200404333A TW200404333A TW092115289A TW92115289A TW200404333A TW 200404333 A TW200404333 A TW 200404333A TW 092115289 A TW092115289 A TW 092115289A TW 92115289 A TW92115289 A TW 92115289A TW 200404333 A TW200404333 A TW 200404333A
- Authority
- TW
- Taiwan
- Prior art keywords
- processing device
- setting data
- data
- correlation
- processing
- Prior art date
Links
- 238000012545 processing Methods 0.000 title claims abstract description 538
- 238000000034 method Methods 0.000 title claims abstract description 160
- 238000010202 multivariate logistic regression analysis Methods 0.000 title claims abstract description 14
- 238000000491 multivariate analysis Methods 0.000 claims description 105
- 238000001514 detection method Methods 0.000 claims description 83
- 238000012360 testing method Methods 0.000 claims description 26
- 238000004458 analytical method Methods 0.000 claims description 21
- 238000004519 manufacturing process Methods 0.000 abstract description 8
- 210000002381 plasma Anatomy 0.000 description 216
- 235000012431 wafers Nutrition 0.000 description 70
- 239000007789 gas Substances 0.000 description 49
- 230000003287 optical effect Effects 0.000 description 22
- 238000004364 calculation method Methods 0.000 description 15
- 238000013500 data storage Methods 0.000 description 15
- 238000012549 training Methods 0.000 description 13
- 238000005530 etching Methods 0.000 description 11
- 238000007689 inspection Methods 0.000 description 9
- 238000005259 measurement Methods 0.000 description 8
- 230000005856 abnormality Effects 0.000 description 7
- 238000003745 diagnosis Methods 0.000 description 7
- 238000002474 experimental method Methods 0.000 description 7
- 239000003990 capacitor Substances 0.000 description 6
- 230000001276 controlling effect Effects 0.000 description 5
- 239000003507 refrigerant Substances 0.000 description 5
- 238000005070 sampling Methods 0.000 description 5
- 230000002159 abnormal effect Effects 0.000 description 4
- 230000005540 biological transmission Effects 0.000 description 4
- 230000007246 mechanism Effects 0.000 description 4
- 239000004065 semiconductor Substances 0.000 description 4
- 238000010586 diagram Methods 0.000 description 3
- 238000009826 distribution Methods 0.000 description 3
- 238000012544 monitoring process Methods 0.000 description 3
- 238000009832 plasma treatment Methods 0.000 description 3
- XUIMIQQOPSSXEZ-UHFFFAOYSA-N Silicon Chemical compound [Si] XUIMIQQOPSSXEZ-UHFFFAOYSA-N 0.000 description 2
- 229910052782 aluminium Inorganic materials 0.000 description 2
- XAGFODPZIPBFFR-UHFFFAOYSA-N aluminium Chemical compound [Al] XAGFODPZIPBFFR-UHFFFAOYSA-N 0.000 description 2
- 230000000875 corresponding effect Effects 0.000 description 2
- 239000013078 crystal Substances 0.000 description 2
- 239000011810 insulating material Substances 0.000 description 2
- 239000000463 material Substances 0.000 description 2
- 229910052710 silicon Inorganic materials 0.000 description 2
- 239000010703 silicon Substances 0.000 description 2
- 239000002002 slurry Substances 0.000 description 2
- 229910052778 Plutonium Inorganic materials 0.000 description 1
- 235000006732 Torreya nucifera Nutrition 0.000 description 1
- 244000111306 Torreya nucifera Species 0.000 description 1
- 229910052770 Uranium Inorganic materials 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 229910052797 bismuth Inorganic materials 0.000 description 1
- JCXGWMGPZLAOME-UHFFFAOYSA-N bismuth atom Chemical compound [Bi] JCXGWMGPZLAOME-UHFFFAOYSA-N 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000001816 cooling Methods 0.000 description 1
- 230000002596 correlated effect Effects 0.000 description 1
- 238000013499 data model Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000005684 electric field Effects 0.000 description 1
- 230000005611 electricity Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 239000000835 fiber Substances 0.000 description 1
- 239000011521 glass Substances 0.000 description 1
- 230000010365 information processing Effects 0.000 description 1
- 238000013101 initial test Methods 0.000 description 1
- 238000004020 luminiscence type Methods 0.000 description 1
- 238000002156 mixing Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 239000013307 optical fiber Substances 0.000 description 1
- 238000001020 plasma etching Methods 0.000 description 1
- OYEHPCDNVJXUIW-UHFFFAOYSA-N plutonium atom Chemical compound [Pu] OYEHPCDNVJXUIW-UHFFFAOYSA-N 0.000 description 1
- 238000010248 power generation Methods 0.000 description 1
- 238000002360 preparation method Methods 0.000 description 1
- 229910052704 radon Inorganic materials 0.000 description 1
- SYUHGPGVQRZVTB-UHFFFAOYSA-N radon atom Chemical compound [Rn] SYUHGPGVQRZVTB-UHFFFAOYSA-N 0.000 description 1
- 239000000523 sample Substances 0.000 description 1
- 230000003595 spectral effect Effects 0.000 description 1
- 238000004611 spectroscopical analysis Methods 0.000 description 1
- 230000000087 stabilizing effect Effects 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
- 239000000758 substrate Substances 0.000 description 1
- 230000007723 transport mechanism Effects 0.000 description 1
- JFALSRSLKYAFGM-UHFFFAOYSA-N uranium(0) Chemical compound [U] JFALSRSLKYAFGM-UHFFFAOYSA-N 0.000 description 1
Classifications
-
- H—ELECTRICITY
- H01—ELECTRIC ELEMENTS
- H01J—ELECTRIC DISCHARGE TUBES OR DISCHARGE LAMPS
- H01J37/00—Discharge tubes with provision for introducing objects or material to be exposed to the discharge, e.g. for the purpose of examination or processing thereof
- H01J37/32—Gas-filled discharge tubes
- H01J37/32917—Plasma diagnostics
- H01J37/32935—Monitoring and controlling tubes by information coming from the object and/or discharge
-
- H—ELECTRICITY
- H01—ELECTRIC ELEMENTS
- H01L—SEMICONDUCTOR DEVICES NOT COVERED BY CLASS H10
- H01L21/00—Processes or apparatus adapted for the manufacture or treatment of semiconductor or solid state devices or of parts thereof
- H01L21/02—Manufacture or treatment of semiconductor devices or of parts thereof
- H01L21/04—Manufacture or treatment of semiconductor devices or of parts thereof the devices having potential barriers, e.g. a PN junction, depletion layer or carrier concentration layer
- H01L21/18—Manufacture or treatment of semiconductor devices or of parts thereof the devices having potential barriers, e.g. a PN junction, depletion layer or carrier concentration layer the devices having semiconductor bodies comprising elements of Group IV of the Periodic Table or AIIIBV compounds with or without impurities, e.g. doping materials
- H01L21/30—Treatment of semiconductor bodies using processes or apparatus not provided for in groups H01L21/20 - H01L21/26
- H01L21/302—Treatment of semiconductor bodies using processes or apparatus not provided for in groups H01L21/20 - H01L21/26 to change their surface-physical characteristics or shape, e.g. etching, polishing, cutting
- H01L21/306—Chemical or electrical treatment, e.g. electrolytic etching
- H01L21/3065—Plasma etching; Reactive-ion etching
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Plasma & Fusion (AREA)
- Computer Hardware Design (AREA)
- General Physics & Mathematics (AREA)
- Manufacturing & Machinery (AREA)
- Condensed Matter Physics & Semiconductors (AREA)
- Microelectronics & Electronic Packaging (AREA)
- Power Engineering (AREA)
- Chemical & Material Sciences (AREA)
- Analytical Chemistry (AREA)
- Drying Of Semiconductors (AREA)
- Plasma Technology (AREA)
Abstract
Description
200404333 (1) 玖、發明說明 【發明所屬之技術領域】 本發明係關於處理裝置之多變量解析模型作成方法、 處理裝置用之多變量解析方法、處理裝置之控制裝置、處 理裝置之控制系統。 【先前技術】 在半導體製造工程中，有使用種種的處理裝置。在半 導體晶圓或玻璃基板等之被處理體的成膜工程或蝕刻工程 中，廣泛使用電漿處理裝置等之處理裝置。各處理裝置分 別具有對於被處理體之固有的製程特性。因此，監視每一 裝置的製程特性，或者預測製程特性等，以進行晶圓的最 佳處理。 例如，在日本專利特開平6- 1 3 22 5 1號公報中，提案 有關於電漿蝕刻裝置的蝕刻監視器。在此情形下，事先硏 究鈾刻的處理結果（均勻性、尺寸精度、形狀或和基底膜 的選擇性等）和電漿的分光分析結果或製程條件（壓力、 氣體流量、偏壓電壓等）的變動狀況等之關係，藉由將這 些記億爲資料庫’㊆夠不直接檢查晶圓而間接地監視處理 結果。在監視結果不符合檢查條件時，將該資訊傳送給触 刻裝置，修正處理條件，或者中止處理，同時對管理者通 報該旨意。 另外，在日本專利特開平10-125660號公報中，提案 關於電漿處理裝置之製程監視方法。在此情形下，在處理 -5- (2) (2)200404333 前，利用試用晶圓，製作使反映電漿狀態的電氣訊號和電 漿處理特性相關連的模型，將實際處理晶圓時所獲得的電 氣訊號的檢測値代入模型，與預測、診斷電漿狀態。 另外，在日本專利特開平1 1 - 8 7 3 2 3號公報中，提案 有關於利用半導體晶圓處理系統的多數參數，以監視製程 之方法及裝置。在此情形下，分析多數的製程參數，使這 些參數統計上相關，以檢測製程特性或者系統特性的變 化。多數的製程參數係使用：發光、環境參數（反應腔內 的壓力或溫度等）、RF功率參數（反射功率、調諧電壓 等）、系統參數（特定的系統構造或控制電壓）。 但是，在習知技術之情形下，藉由多變量解析等之統 計手法而解析種種的測量資料以作成模型，利用此模型以 掌握、監視處理裝置的狀態或製程特性，例如，在附設於 各處理裝置之感測器間的個體差等，每一處理裝置在製程 特性上有差異時，即使就一個之處理裝置作成模型，也無 法將此模型運用於同統之其他的處理裝置，必須每一處理 裝置取得種種的測量資料，而每次作成模型，存在有在模 型的作成上需要多的工夫和時間之課題。另外，在製程條 件改變時，也需要每一個製程條件取得種種的測量資料， 每次作成模型，存在有在模型的作成上需要多的工夫和時 間之課題。 本發明係爲了解決上述課題所完成者，目的在於提 供：每一處理裝置即使在製程特性或製程條件有差異，如 就一個處理裝置作成模型，也可以將該模型援用於同種的 -6 - (3) (3)200404333 其他處理裝置，能夠減輕每一處理裝置在作成模型時的工 夫或負擔，另外，即使不每一處理裝置新作成模型，也可 以評估各處理裝置之裝置狀態的處理裝置之多變量解析模 型作成方法及處理裝置用之多變量解析方法。 【發明內容】 爲了解決上述課題，如依據本發明之第1觀點，係提 供一種處理裝置之多變量解析模型作成方法，是針對作成 藉由多變量解析，以評估處理裝置之裝置狀態，或者預測 處理結果時的多變量解析模型之方法，其特徵爲具有：在 多數的處理裝置中，在個別依據第1設定資料而動作時， 藉由多變量解析而就上述每一個處理裝置求得由上述各處 理裝置的多數感測器所檢測的檢測資料和上述第1設定資 料的相關關係的第1工程；及如將在上述各處理裝置中的 1個當成基準處理裝置時，在此基準處理裝置中，於依據 新的第2設定資料而動作時，藉由多變量解析以求得由上 述基準處理裝置的多數感測器所檢測的檢測資料和上述第 2設定資料的相關關係的第2工程；及依據在上述第1工 程所求得的上述其他處理裝置的相關關係，和在上述第！ 工程所求得的上述基準處理裝置的相關關係，和在上述第 2工程所求得的上述基準處理裝置的上述相關關係，以求 得上述基準處理裝置以外的其他處理裝置的上述第2設定 資料和檢測資料的相關關係，作成依據如此求得的相關關 係’以評估上述其他處理裝置的裝置狀態或者預測處理結 (4) (4)200404333 果的多變量解析模型的第3工程。 爲了解決上述課題，如依據本發明之第2觀點，係提 供一種處理裝置用之多變量解析模型作成方法，是針對作 成藉由多變量解析，以評估處理裝置之裝置狀態，或者預 測處理結果時的多變量解析方法，其特徵爲具有：在多數 的處理裝置中，在個別依據第1設定資料而動作時，藉由 多變量解析而就上述每一個處理裝置求得由上述各處理裝 置的多數感測器所檢測的檢測資料和上述第1設定資料的 相關關係的第1工程；及如將在上述各處理裝置中的1個 當成基準處理裝置時，在此基準處理裝置中，於依據新的 第2設定資料而動作時，藉由多變量解析以求得由上述基 準處理裝置的多數感測器所檢測的檢測資料和上述第2設 定資料的相關關係的第2工程；及依據在上述第1工程所 求得的上述其他處理裝置的相關關係，和在上述第1工程 所求得的上述基準處理裝置的相關關係，和在上述第2工 程所求得的上述基準處理裝置的上述相關關係，以求得上 述基準處理裝置以外的其他處理裝置的上述第2設定資料 和檢測資料的相關關係，作成依據如此求得的相關關係， 以評估上述其他處理裝置的裝置狀態或者預測處理結果的 多變量解析模型的第3工程。 另外，在依據上述第1觀點及第2觀點的發明中，上 述第3工程也可以依據對於在上述第1工程所求得的上述 其他處理裝置的相關關係的上述其他處理裝置的上述第2 設定資料和檢測資料的相關關係，和對於在上述第1工程 -8- (5) (5)200404333 所求得的上述基準處理裝置的相關關係的在上述第2工程 所求得的上述基準處理裝置的上述相關關係之比例關係， 以求得上述其他處理裝置的上述第2設定資料和檢測資料 的相關關係。另外，上述多變量解析例如也可以藉由部份 最小平方法（PLS法）進行。 另外，在依據上述第1觀點及第2觀點的發明中，處 理裝置可以爲電漿處理裝置。在此情形下，上述設定資料 係使用可以控制電漿狀態的多數之控制參數，同時，上述 檢測資料可以使用由反映電漿狀態的多數之電漿反映參 數、與裝置狀態相關的多數之裝置狀態參數、反映製程完 成之參數群中所選擇的至少其中1種或者2種以上的參 數。 另外，在依據上述第2觀點的發明中，上述多變量解 析模型可以爲由在上述第3工程所求得的上述其他處理裝 置之相關關係和上述第2設定資料所算出的檢測資料和上 述第2設定資料的相關關係式。 爲了解決上述課題，如依據本發明之第3觀點，係提 供一種處理裝置之控制裝置，是針對設置在處理被處理體 的處理裝置，依據特定的設定資料，以進行上述處理裝置 的控制之處理裝置之控制裝置，其特徵爲：設置連接於上 述處理裝置及至少成爲基準的處理裝置及主機裝置相連接 的網路，可以進行資料之交換的發送接收手段，在基於第 1設定資料而動作時，藉由上述發送接收手段，將由上述 處理裝置的多數感測器所檢測出的檢測資料和上述第1設 -9 - (6) (6)200404333 定資料透過上述網路而發送給上述主機裝置，將依據所發 送的資料，藉由上述主機裝置以多變量解析所求得之上述 第1設定資料和上述檢測資料的相關關係由上述主機裝置 藉由上述發送接收手段而透過網路予以接收，藉由上述發 送接收手段將新的第2設定資料透過上述網路發送給主機 裝置，將依據所發送的資料，藉由上述主機裝置所求得的 上述第2設定資料和基於此第2設定資料之檢測資料的相 關關係由上述主機裝置藉由上述發送接收手段透過上述網 路而予以接收，依據由上述主機裝置所接收的上述第2設 定資料的相關關係，作成多變量解析模型，依據此多變量 解析模型，以評估上述處理裝置之裝置狀態或者預測處理 結果，因應該結果以控制上述處理裝置。 另外，在依據上述第3觀點的發明中，可以使上述檢 測資料算出手段計算藉由上述發送接收手段透過上述網路 接收作成在以上述其他處理裝置進行特定製程處理時，評 估裝置狀態或者預測處理結果之多變量解析模型用的其他 處理裝置的設定資料，由接收的上述設定資料和上述處理 裝置的上述相關關係，以與上述其他的處理裝置的上述特 定製程處理相同的條件使上述處理裝置動作時的上述處理 裝置之檢測資料。 另外，在依據上述第3觀點的發明中，上述其他處理 裝置的設定資料可以利用：在上述特定製程處理前，藉由 多變量解析所求得之上述其他處理裝置的設定資料，和由 基於此設定資料而動作時的上述其他處理裝置之多數感測 •10- (7) (7)200404333 器所檢測的檢測資料的相關關係，以及由上述其他處理裝 置在進行上述特定製程處理時的上述其他處理裝置之多數 感測器所檢測的檢測資料而被算出。 另外，在依據上述第3觀點的發明中，關於上述處理 裝置的上述第2設定資料的相關關係也可以依據：由上述 主機裝置藉由多變量解析所求得之關於上述處理裝置的上 述第1設定資料的相關關係，和由上述主機裝置藉由多變 量解析所求得之上述基準處理裝置依據第1設定資料而動 作時，由上述基準處理裝置的多數感測器所檢測的檢測資 料與上述第1設定資料的相關關係，和由上述主機裝置藉 由多變量解析所求得之上述基準處理裝置依據新的第2設 定資料而動作時，由上述基準處理裝置的多數感測器所檢 測的檢測資料與上述第2設定資料的相關關係而藉由上述 主機裝置所算出。 另外，在基於上述第3觀點的發明中，處理裝置也可 以爲電漿處理裝置。在此情形下，上述設定資料係使用可 以控制電漿狀態的多數之控制參數，同時，上述檢測資料 可以使用由反映電漿狀態的多數之電漿反映參數、與裝置 狀態相關的多數之裝置狀態參數、反映製程完成之參數群 中所選擇的至少其中1種或者2種以上的參數。另外，上 述多變量解析可藉由部份最小平方法進行。另外，上述處 理裝置也可以爲電漿處理裝置。 爲了解決上述課題，如依據本發明之第4觀點，係提 供一種處理裝置之控制系統，是針對具備依據特定的設定 -11 - (8) (8)200404333 資料，以進行處理被處理體之處理裝置的控制之控制裝置 的處理裝置之控制系統，其特徵爲：具備透過發送接收手 段被連接於網路的多數之前述處理裝置，及連接在上述網 路的主機裝置，上述主機裝置在多數的處理裝置分別依據 第1設定資料動作時，一由上述多數的處理裝置透過上述 網路接收由上述各處理裝置的多數感測器所檢測的檢測資 料和上述第1設定資料時，便藉由多變量解析而每一上述 各處理裝置求得接收的上述第1設定資料和上述檢測資料 的相關關係，將求得之相關關係透過上述網路而發送給對 應的處理裝置，上述主機裝置在上述各處理裝置中的當成 基準之處理裝置依據新的第2設定資料動作時，一由上述 基準處理裝置透過上述網路接收由上述基準處理裝置的多 數感測器所檢測的檢測資料和上述第2設定資料時，便藉 由多變量解析以求得接收的上述第1設定資料和上述檢測 資料的相關關係，將求得的相關關係透過上述網路而發送 給上述基準處理裝置，上述主機裝置一透過上述網路由上 述基準處理裝置以外的其他處理裝置接收上述第2設定資 料時，便依據藉由多變量解析所求得之關於上述其他處理 裝置的上述第1設定資料的上述相關關係，和藉由上述多 變量解析所求得之關於上述基準處理裝置的上述第1設定 資料之上述相關關係，和藉由上述多變量解析所求得之關 於上述基準處理裝置的上述第2設定資料之上述相關關 係，以求得接收的上述第2設定資料和基於此第2設定資 料之檢測資料的相關關係，將所求得之相關關係透過上述 -12- (9) (9)200404333 網路而發送給上述其他處理裝置，上述其他處理裝置依據 由上述主機裝置所接收的關於上述第2設定資料的相關關 係，以作成多變量解析模型，依據此多變量解析模型，以 評估上述處理裝置的裝置狀態或者預測處理結果，因應該 結果，以控制上述處理裝置。 另外，在依據上述第4觀點的發明中，處理裝置可以 爲電榮處理裝置。在此情形下，上述設定資料係使用可以 控制電漿狀態的多數之控制參數，同時，上述檢測資料可 以使用由反映電漿狀態的多數之電漿反映參數、與裝置狀 態相關的多數之裝置狀態參數、反映製程完成之參數群中 所選擇的至少其中1種或者2種以上的參數。另外，上述 多變量解析可藉由部份最小平方法進行。另外，上述處理 裝置也可以爲電漿處理裝置。 【實施方式】 以下一面參考所附圖面，一面詳細說明關於本發明之 裝置的合適實施形態。另外，在本說明書及圖面中，關於 實質上具有相同的機能構造之構成要素，藉由賦予相同的 圖號，省略重複說明。 首先，參考第1圖、第2圖以說明關於本發明之第1 實施形態的電漿處理裝置。本實施形態之電漿處理裝置 100係如第1圖所示般，具備：鋁製的處理室（處理腔） 101，及透過絕緣材102A以支持配置在此處理室101內 的下部電極102而可以升降之鋁製支持體103，及配置在 -13- (10) (10)200404333 此支持體103的上方’且供給製程氣體，而且兼爲上部電 極的淋浴頭（以下’需要時，也稱爲「上部電極」） 104° 上述處理室101係形成爲上述爲小直徑的上室 10 1A，下部形成爲大直徑的下室101B。上室101A係由 偶極子環型磁鐵1 〇 5所包圍。此偶極子環型磁鐵丨〇 5係在 由環型的磁性體所形成的機殼內收容多數的非等向性弧形 柱狀磁鐵而形成，在上室101A內形成整體爲朝向一個方 向的一樣的水平磁場。在下室1 0 1 B之上部形成搬入搬出 晶圓W用之出入口’在此出入口安裝閘門閥106。 在下部電極1〇2透過匹配器7A而連接高頻電源 107，由此高頻電源107對於下部電極1〇2施加13.56 Μ 之高頻電力Ρ，在上室101Α內，與上部電極104之間形 成垂直方向的電場。此高頻電力Ρ係藉由連接在高頻電源 107和匹配器7Α間的瓦特計1〇7而檢測出。此高頻電力 Ρ爲可以控制的參數，在本實施形態中，將高頻電力Ρ與 後述的氣體流量、電極間距離等之可控制參數一同定義爲 控制參數。另外，控制參數係對於電漿處理裝置爲可以設 定之參數，所以也稱爲設定資料。 在上述匹配器7Α之下部電極102側（高頻電壓的輸 出側）安裝電氣量測器（例如，VI探針）107C，藉由透 過此電氣量測器107C而施加在下部電極102的高頻電力 Ρ，將基於發生在上室101Α內之電漿的基本波及高次諧 波的高頻電壓V、高頻電流I、電壓波形和電流波形間的 -14- (11) (11)200404333 相位差P當成電氣資料予以檢測。這些電氣資料係與後述 的光學資料一齊地反映電漿狀態之可監視的參數，在本實 施形態中，定義爲電漿反映參數。另外，電漿反映參數係 藉由電氣量測器1 0 7 C所檢測出之資料，所以也稱爲檢測 資料。 上述匹配器7A例如內藏2個可變電容器Cl、C2、 電容器C及線圈L，透過可變電容器C1、C 2可以取得阻 抗匹配。在匹配狀態的可變電容器Cl、C2之電容、由上 述匹配器7 A內之量測器（未圖示出）所測量之高頻電壓 Vpp 係與後述的 APC(Automatic pressure controller :自動 壓力控制器）開度等一同爲顯示處理時的裝置狀態之參 數，在本實施形態中，將顯示裝置狀態之可變電容器 Cl、C2之電容、高頻電壓Vpp及APC之開度分別定義爲 裝置狀態參數。然後，裝置狀態參數係無法控制的參數， 爲可以檢測之資料，也稱爲檢測資料。 在上述下部電極102的上面配置靜電夾頭108，在此 靜電夾頭108之電極板108A連接直流電源1〇9。因此， 在高真空下，藉由直流電源109對於電極板i〇8A施加高 電壓，以靜電夾頭1 〇 8靜電吸引晶圓 W。在此下部電極 102的外圍配置聚焦環110，將在上室101A內所產生的 電漿聚集在晶圓W。另外，在聚焦環1 1 〇之下側配置安裝 於支持體1 03之上部的排氣環1 1 1。在此排氣環1 1 1涵蓋 全周在圓周方向等間隔形成多數孔，透過這些孔，將上室 101A內之氣體排往下室101B。 -15- (12) (12)200404333 上述支持體103係透過滾珠導螺桿機構1 12及波紋管 113而可在上室101A和下室101B間升降。因此，在將晶 圓W供應給下部電極1〇2上時，下部電極102透過支持 體1 0 3而下降至下室1 〇 1 B，開放閘門閥1 0 6透過未圖示 出的搬運機構而將晶圓W供應給下部電極1 02上。下部 電極102和上部電極1〇4之間的電極間距離係可設定爲特 定値之參數，如上述般，構成爲控制參數。 在支持體103的內部形成連接於冷媒配管1 14之冷媒 流路103A，透過冷媒配管i 14在冷媒流路103A內使冷媒 循環，將晶圓W調整爲特定溫度。在支持體1 〇3、絕緣材 102A、下部電極102及靜電夾頭108分別形成氣體流路 103B，由氣體導入機構115透過氣體配管n5A將He氣 體當成背面氣體以特定壓力而供應於靜電夾頭108和晶圓 W之間的間隙，透過H e氣體，提高靜電夾頭1 0 8和晶圓 W間的熱傳導性。另外，i〗6係波紋管蓋。 在上述淋浴頭104之上面形成氣體導入部104A，在 此氣體導入部1 04 A透過配管1 1 7連接製程氣體供給系統 1 1 8。製程氣體供給系統〗丨8系具有Ar氣體供給源 1 1 8 A、C Ο r氣體供給源1 1 8 B、C 4 F 6氣體供給源1 1 8 C及 〇2氣體供給源1 18D。這些氣體供給源1 18A、1 18B、 118C、118D 分別透過閥門 H8E、118F、118G、118H 及 質流控制器1 1 8 I、1 1 8 J、1 1 8 K、1 1 8 L以特定之設定流量 將個別的氣體供應給淋浴頭1 04，在其內部調整爲具有特 定的混合比之混合氣體。各氣體流量爲藉由個別之質流控 -16 - (13) (13)200404333 制器1 1 81、1 1 8J、1 18K、1 18L而可以控制，且可以檢測 之參數，如上述般，構成爲控制參數。 在上述淋浴頭104之下面涵蓋全面而均等配置多數的 孔1 04B，透過這些孔1 〇4B由淋浴頭104將混合氣體當成 製程氣體供應給上室101A內。另外，在下室101B之下 部的排氣孔連接排氣管1 0 1 C，透過由連接於此排氣管 101C之真空泵等所形成的排氣系統119而排氣處理室101 內以保持特定的氣體壓力。在排氣管1 0 1 C設置APC閥門 101D，依循處理室1〇1內之氣體壓力而自動調節開度。 此開度爲顯示裝置狀態的裝置狀態參數，是無法控制的參 數。 在上述處理室1 0 1的側壁設置檢測窗1 2 1，在處理室 1 0 1的側壁外側透上述檢測窗1 2 1將處理室1 〇 1內之電漿 發光橫跨多波長予以檢測之分光器（以下，稱爲「光學檢 測器」）1 2 0。依據關於藉由此光學檢測器1 2 0所獲得之 特定波長的光學資料以監視電漿狀態，例如，檢測電漿處 理之終點。此光學資料係與基於藉由高頻電力P所發生的 電漿的電氣資料一齊地構成反映電獎狀態的電獎反映參 數。 接著，一面參考圖面一面說明設置在上述電漿處理裝 置100的多變量解析手段。電漿處理裝置100例如具備如 第2圖所示之多變量解析手段200。此多變量解析手段 2 0 0爲具備：記憶多變量解析程式的多變量解析程式記憶 手段20 1 ;及間歇地取樣由控制參數量測器22丨、電漿反 -17- (14) (14)200404333 映參數量測器222及裝置狀態參數量測器22 3來之檢測訊 號的控制參數訊號取樣手段202、電漿反映參數訊號取樣 手段203及裝置狀態參數訊號取樣手段204。另外，具 備：記憶使多數的電漿反映參數（電氣資料及光學資 料）、與裝置狀態相關的多數的裝置狀態參數及多數的控 制參數相關連之模型等的解析結果或解析所必要的資料之 解析資料記憶手段205 ;及透過模型而依據目的以運算控 制參數、電漿反映參數及裝置狀態參數之運算手段206 ; 及依據來自運算手段2 06的運算訊號，依據目的而進行控 制參數、多數的電漿反映參數及裝置狀態參數之預測、診 斷、控制的預測·診斷·控制手段2 0 7。 另外’在多變量解析手段200分別連接依據控制參數 以控制電漿處理裝置100之處理裝置控制手段22 5、警報 器226及顯示手段224。處理裝置控制手段225例如系依 據來自預測·診斷·控制手段207之訊號以繼續或者中斷 晶圓W之處理。警報器226及顯示手段224係如後述 般，依據來自預測·診斷·控制手段2 0 7的訊號，基於目 的以通報控制參數、多數的電漿反映參數及裝置狀態參數 之何者之異常。另外，解析資料記憶手段205係記憶關於 上述各參數之資料或這些之加工資料（使用於多變量解析 之加工資料）。另外，控制參數量測器22 1、電漿反映參 數量測器222、裝置狀態參數量測器223係分別將流量檢 測器、光學量測器、高頻電壓Vpp量測器等之多數的控 制參數之量測器、多數的電漿反映參數的量測器、多數的 -18- (15) (15)200404333 裝置狀態參數的量測器彙整爲一個而顯示。 此處，說明本發明之原理。例如，作爲作成新模型時 的基準之處理裝置，係考慮電漿處理裝置100A，作爲此 基準處理裝置以外的處理裝置係考慮電漿處理裝置 100B。在電漿處理裝置100A、100B之間，由於製造上之 偏差等，存在稍微之個體差。另外，上述電氣量測器 107C、光學檢測器120等之感測器也分別由於製造誤差 等，而每一個電漿處理裝置存在個體差，所以即使在同一 種的電漿處理裝置使用同一種之感測器，也無法獲得相同 的檢測資料。因此，即使爲同一種之電漿處理裝置，需要 每一個電漿處理裝置作成多變量解析模型，無法將一個多 變量解析模型援用爲其他同種之電漿處理裝置的多變量解 析模型。 因此，在本實施形態中，例如，在電漿處理裝置 100A > 100B間，即使有製造上的個體差，或者在個別的 多數感測器間有個體差，也可以將關於電漿處理裝置 100A所作成的多變量解析模型源用於其他的電漿處理裝 置1 00B。在本實施形態中，多變量解析之一手法係利用 部份最小平方法（以下，稱爲「PLS(Partial Least Squares)法」）以作成電漿處理裝置100A、100B個別的 多變量解析模型，找出裝置間的個體差，作成吸收此個體 差的模型。PLS法之詳細例如揭露在 JOURNAL OF CHEMOMETRICS，VOL. 2 (PP.2 1 1 -228)( 1 998)。 例如，電漿處理裝置100A、100B都設多數的控制參 -19- (16) 200404333 數（設定資料）爲目的變數，設多數的電漿反映 電氣資料及光學資料的檢測資料）爲說明變數， 目的變數爲成分之行列X和以說明變數爲成分 相關連的下述（1 )所示之回歸式（以下，單單 型」）（第1工程）。 在電漿處理裝置100A、100B個別的運算 中，依據利用多變量解析之一手法的PLS法， 實驗所獲得之說明變數和目的變數，分別算出模 行列K、Kb，如上述般，將這些模型記憶在解析 手段205。另外，在下述（1) 、（2)的模型中 爲個別之模型的回歸行列，a係表示電漿處理裝ί b係表示電漿處理裝置100Β。200404333 (1) 发明. Description of the invention [Technical field to which the invention belongs] The present invention relates to a method for creating a multivariate analysis model of a processing device, a multivariate analysis method for a processing device, a control device for a processing device, and a control system for a processing device. [Prior Art] There are various processing devices used in semiconductor manufacturing processes. In a film formation process or an etching process of a processing object such as a semiconductor wafer or a glass substrate, a processing apparatus such as a plasma processing apparatus is widely used. Each processing device has process characteristics inherent to the object to be processed. Therefore, monitor the process characteristics of each device, or predict the process characteristics, etc., for optimal wafer processing. For example, in Japanese Patent Application Laid-Open No. 6-123222, an etching monitor for a plasma etching apparatus is proposed. In this case, study the processing results (uniformity, dimensional accuracy, shape, or selectivity of the base film, etc.) of the uranium etching and the plasma spectroscopic analysis results or process conditions (pressure, gas flow rate, bias voltage, etc.) in advance. The relationship between the changes in the status of), etc., by recording these billions as a database, it is sufficient to indirectly monitor the processing results without directly inspecting the wafer. When the monitoring result does not meet the inspection conditions, the information is transmitted to the touch device, the processing conditions are corrected, or the processing is suspended, and the manager is notified of the intention. In addition, Japanese Patent Laid-Open No. 10-125660 proposes a process monitoring method for a plasma processing apparatus. In this case, before processing -5- (2) (2) 200404333, use the trial wafer to create a model that correlates the electrical signal reflecting the state of the plasma with the plasma processing characteristics. The detection of the obtained electrical signals is substituted into the model, and the plasma state is predicted and diagnosed. In addition, Japanese Patent Laid-Open No. 1 1-8 7 3 2 3 proposes a method and an apparatus for monitoring a process by using most parameters of a semiconductor wafer processing system. In this case, most process parameters are analyzed and these parameters are statistically correlated to detect changes in process characteristics or system characteristics. Most process parameters are used: luminescence, environmental parameters (pressure or temperature in the reaction chamber, etc.), RF power parameters (reflected power, tuning voltage, etc.), system parameters (specific system construction or control voltage). However, in the case of conventional technology, various measurement data are analyzed by statistical methods such as multivariate analysis to create a model, and this model is used to grasp and monitor the status or process characteristics of the processing device. For example, Individual differences among sensors of processing devices, etc. When the processing characteristics of each processing device are different, even if one processing device is modeled, this model cannot be applied to other processing devices of the same system. A processing device acquires various measurement data, and each time a model is created, there is a problem that it takes much time and time to create the model. In addition, when the process conditions are changed, it is also necessary to obtain various measurement data for each process condition. Each time a model is created, there is a problem that it takes much time and time to create a model. The present invention has been completed in order to solve the above-mentioned problems, and the object is to provide: even if each processing device has a difference in process characteristics or processing conditions, if a model is created for one processing device, the model can be used for the same type of -6-( 3) (3) 200404333 Other processing devices can reduce the time or burden of each processing device when creating a model. In addition, even if a new model is not created for each processing device, the status of each processing device can be evaluated. Multivariate analysis model creation method and multivariate analysis method used by processing device. [Summary of the Invention] In order to solve the above-mentioned problems, according to the first aspect of the present invention, a method for creating a multivariate analysis model of a processing device is provided, which aims at evaluating the state of the device of the processing device through multivariate analysis, or predicting The method of multivariate analysis model for processing results is characterized in that: in most processing apparatuses, when operating individually based on the first setting data, multivariate analysis is used to obtain each of the processing apparatuses from the above. The first process of the correlation between the detection data detected by the plurality of sensors of each processing device and the first setting data; and if one of the processing devices is used as a reference processing device, the reference processing device is used here. In the second process, when operating based on the new second setting data, a multivariate analysis is performed to obtain the correlation between the detection data detected by the plurality of sensors of the reference processing device and the second setting data. ; And the relationship between the other processing devices based on the above-mentioned first process, and in the above-mentioned first! The correlation between the reference processing device obtained in the project and the correlation between the reference processing device obtained in the second project to obtain the second setting data of the processing device other than the reference processing device. Correlation with test data, based on the correlation obtained in this way, to evaluate the device state of the other processing devices mentioned above or predict the processing results of the multivariate analysis model (4) (4) 200404333. In order to solve the above-mentioned problems, according to the second aspect of the present invention, a method for creating a multivariate analysis model for a processing device is provided. The method is aimed at evaluating the device state of the processing device through multivariate analysis or predicting the processing result. The multivariate analysis method is characterized in that when a plurality of processing devices operate individually based on the first setting data, a multivariate analysis is used to obtain a majority for each of the processing devices. The first process of the correlation between the detection data detected by the sensor and the first setting data; and if one of the processing devices described above is used as a reference processing device, in this reference processing device, based on the new When the second set of data is operated, the second process of obtaining the correlation between the detection data detected by the plurality of sensors of the reference processing device and the second set of data by multivariate analysis; Correlation between the other processing devices obtained in the first process and the reference processing device obtained in the first process The correlation relationship and the correlation relationship between the reference processing device obtained in the second project to obtain the correlation between the second setting data and detection data of a processing device other than the reference processing device, and the basis is made as follows The obtained correlations are used to evaluate the third state of the above-mentioned other processing devices or the multivariate analysis model for predicting processing results. In addition, in the invention according to the first and second aspects, the third process may be based on the second setting of the other processing apparatus related to the relationship between the other processing apparatus obtained in the first process. The correlation between the data and the test data, and the correlation between the reference processing device obtained in the second process and the reference processing device obtained in the first process-8- (5) (5) 200404333 To obtain the correlation between the second setting data and the detection data of the other processing devices. The multivariate analysis described above can be performed by, for example, the partial least square method (PLS method). In the invention according to the first and second aspects, the processing device may be a plasma processing device. In this case, the above-mentioned setting data uses the majority of control parameters that can control the state of the plasma, and the above-mentioned detection data can use the majority of the device states related to the state of the device that reflect the parameters of the plasma that reflect the state of the plasma. At least one or two or more parameters selected in the parameter group reflecting the completion of the process. Moreover, in the invention according to the second aspect, the multivariate analysis model may be the detection data calculated from the correlation between the other processing devices obtained in the third process and the second setting data, and the first 2 Set the correlation of the data. In order to solve the above-mentioned problems, according to the third aspect of the present invention, a control device for a processing device is provided, which is directed to a processing device installed on a processing object, and performs processing for controlling the processing device according to specific setting data. The control device of the device is characterized in that: a network connected to the processing device, at least a reference processing device, and a host device is provided, and a sending and receiving means capable of data exchange can be performed when the device operates based on the first setting data By means of the sending and receiving means, the detection data detected by the majority of sensors of the processing device and the first setting -9-(6) (6) 200404333 are sent to the host device through the network. The correlation between the first setting data and the detection data obtained by the host device through multivariate analysis based on the sent data will be received by the host device through the network through the sending and receiving means, Send the new second setting data to the host device via the network by the sending and receiving means , According to the sent data, the correlation between the second setting data obtained by the host device and the detection data based on the second setting data will be determined by the host device through the network through the sending and receiving means. Receive it, and make a multivariate analysis model based on the correlation between the second setting data received by the host device, and use this multivariate analysis model to evaluate the device state of the processing device or predict the processing result. Controlling the processing device. In addition, in the invention according to the third aspect, the detection data calculation means may be calculated and received through the network by the transmission and reception means, and the state of the apparatus or the prediction process may be evaluated when the specific process is performed by the other processing apparatus. The setting data of the other processing device for the multivariate analysis model of the result is to operate the processing device under the same conditions as the specific process processing of the other processing device based on the received setting data and the correlation between the processing device. Test data of the above-mentioned processing device at the time. In addition, in the invention according to the third aspect, the setting data of the other processing device may be used: before the specific process is processed, the setting data of the other processing device obtained through multivariate analysis, and Most of the above-mentioned other processing devices sense when the data is set to operate • 10- (7) (7) 200404333 Correlation of the detection data detected by the device, and the above-mentioned other processing devices when the above-mentioned other processing device performs the specific process processing described above The detection data detected by most sensors of the processing device is calculated. In addition, in the invention according to the third aspect, the correlation between the second setting data of the processing device may be based on the first of the processing device obtained by the host device through multivariate analysis. When the correlation between the setting data and the reference processing device obtained by the host device through multivariate analysis operates according to the first setting data, the detection data detected by the majority of sensors of the reference processing device and the above When the correlation between the first setting data and the reference processing device obtained by the host device through multivariate analysis operates according to the new second setting data, it is detected by the majority of sensors of the reference processing device. The correlation between the detection data and the second setting data is calculated by the host device. In the invention according to the third aspect, the processing device may be a plasma processing device. In this case, the above-mentioned setting data uses the majority of control parameters that can control the state of the plasma, and the above-mentioned detection data can use the majority of the device states related to the state of the device that reflect the parameters of the plasma that reflect the state of the plasma. At least one or two or more parameters selected in the parameter group reflecting the completion of the process. In addition, the above multivariate analysis can be performed by a partial least square method. The processing device may be a plasma processing device. In order to solve the above-mentioned problems, according to the fourth aspect of the present invention, a control system for a processing device is provided, which is provided with data according to a specific setting -11-(8) (8) 200404333 to process the object to be processed. The control system of the processing device of the control device of the device is characterized in that it includes a plurality of the aforementioned processing devices connected to the network by means of transmission and reception, and a host device connected to the network. When the processing device operates according to the first setting data, one of the plurality of processing devices receives the detection data detected by the plurality of sensors of the processing devices and the first setting data through the network. The variables are analyzed, and each of the processing devices obtains the correlation between the received first setting data and the detection data, and sends the obtained correlation to the corresponding processing device through the network. The host device When the processing device serving as a reference in the processing device operates according to the new second setting data, When the processing device receives the detection data and the second setting data detected by the plurality of sensors of the reference processing device through the network, the multi-variable analysis is used to obtain the received first setting data and the detection data. The obtained correlation is sent to the reference processing device through the network. When the host device receives the second setting data through a route to another processing device other than the reference processing device through the network, the host device is based on the borrowed information. The above-mentioned correlation between the first setting data of the other processing device obtained by the multivariate analysis and the above-mentioned correlation between the first setting data of the reference processing device obtained by the multivariate analysis , And the above-mentioned correlation between the second setting data on the reference processing device obtained by the multivariate analysis, to obtain the correlation between the received second setting data and the detection data based on the second setting data Relationship, the related relationship obtained through the above -12- (9) (9) 200404333 network For the other processing device, the other processing device creates a multivariate analysis model based on the correlation relationship of the second setting data received by the host device, and evaluates the device state of the processing device based on the multivariate analysis model. Alternatively, the processing result is predicted, and the processing device is controlled according to the result. In the invention according to the fourth aspect, the processing device may be a galvanic processing device. In this case, the above-mentioned setting data uses the majority of control parameters that can control the state of the plasma, and the above-mentioned detection data can use the majority of the device states related to the state of the device that reflect the parameters of the plasma that reflect the state of the plasma. At least one or two or more parameters selected in the parameter group reflecting the completion of the process. In addition, the above multivariate analysis can be performed by a partial least square method. The processing apparatus may be a plasma processing apparatus. [Embodiment] A suitable embodiment of the device according to the present invention will be described in detail below with reference to the accompanying drawings. In this specification and the drawings, components having substantially the same functional structure are given the same drawing numbers, and redundant descriptions are omitted. First, a plasma processing apparatus according to a first embodiment of the present invention will be described with reference to FIGS. 1 and 2. As shown in FIG. 1, the plasma processing apparatus 100 of this embodiment includes a processing chamber (processing chamber) 101 made of aluminum, and a lower electrode 102 disposed in the processing chamber 101 through an insulating material 102A. An aluminum support 103 that can be raised and lowered, and a shower head that is arranged above -13- (10) (10) 200404333, and which supplies process gas and also serves as an upper electrode (hereinafter also referred to as "when needed") “The upper electrode”) 104 ° The processing chamber 101 is formed as the upper chamber 101A having a small diameter as described above, and the lower chamber 101B is formed as a large diameter on the lower side. The upper chamber 101A is surrounded by a dipole ring magnet 105. This dipole ring magnet is formed by housing a large number of anisotropic arc-shaped columnar magnets in a casing formed by a ring-shaped magnetic body. The entire shape is formed in the upper chamber 101A so as to face one direction. Same horizontal magnetic field. An inlet / outlet for loading / unloading wafers W is formed on the upper part of the lower chamber 1 0 B. A gate valve 106 is installed at the inlet / outlet. A high-frequency power source 107 is connected to the lower electrode 10 through a matching device 7A, and thus the high-frequency power source 107 applies a high-frequency power P of 13.56 M to the lower electrode 102 between the upper electrode 101A and the upper electrode 104. A vertical electric field is formed. This high-frequency power P is detected by a wattmeter 107 connected between the high-frequency power source 107 and the matching unit 7A. This high-frequency power P is a controllable parameter. In this embodiment, the high-frequency power P is defined as a control parameter together with controllable parameters such as a gas flow rate and a distance between electrodes described later. In addition, the control parameter is a parameter that can be set for the plasma processing apparatus, so it is also called a setting data. An electric measuring device (for example, a VI probe) 107C is installed on the lower electrode 102 side (the high-frequency voltage output side) of the matching device 7A, and the high frequency applied to the lower electrode 102 is transmitted through the electric measuring device 107C. The electric power P will be based on the high-frequency voltage V, the high-frequency current I, the voltage waveform, and the current waveform of -14- (11) (11) 200404333 phase based on the fundamental wave and harmonics of the plasma generated in the upper chamber 101A. The difference P is detected as electrical data. These electrical data are monitorable parameters that reflect the state of the plasma together with the optical data described later. In this embodiment, they are defined as the plasma reflection parameters. In addition, the plasma reflection parameter is the data detected by the electrical measuring device 10 7 C, so it is also called detection data. The matching device 7A includes, for example, two variable capacitors Cl, C2, capacitor C, and coil L, and impedance matching can be achieved through the variable capacitors C1 and C2. The capacitances of the variable capacitors Cl and C2 in the matched state, and the high-frequency voltage Vpp measured by a measuring device (not shown) in the above-mentioned matcher 7 A are the same as those of the APC (Automatic pressure controller) The opening degree and other parameters are also parameters of the device state during display processing. In this embodiment, the capacitances of the variable capacitors Cl, C2, the high-frequency voltage Vpp, and the APC opening degree of the display device state are respectively defined as the device states. parameter. Then, the device state parameter is a parameter that cannot be controlled, and is data that can be detected, also called detection data. An electrostatic chuck 108 is disposed on the lower electrode 102, and an electrode plate 108A of the electrostatic chuck 108 is connected to a DC power source 109. Therefore, under a high vacuum, a high voltage is applied to the electrode plate 108 by the DC power source 109, and the wafer W is electrostatically attracted by the electrostatic chuck 108. A focus ring 110 is arranged on the periphery of the lower electrode 102, and the plasma generated in the upper chamber 101A is collected on the wafer W. Further, an exhaust ring 1 1 1 mounted on an upper portion of the support 10 03 is disposed below the focus ring 1 10. Here, the exhaust ring 1 1 1 covers a plurality of holes formed at regular intervals in the circumferential direction throughout the circumference, and the gas in the upper chamber 101A is exhausted to the lower chamber 101B through the holes. -15- (12) (12) 200404333 The above-mentioned support body 103 can be raised and lowered between the upper chamber 101A and the lower chamber 101B through a ball screw mechanism 112 and a bellows 113. Therefore, when the wafer W is supplied to the lower electrode 102, the lower electrode 102 is lowered to the lower chamber 10b through the support 103, and the gate valve 106 is opened through a transport mechanism (not shown). The wafer W is supplied to the lower electrode 102. The inter-electrode distance between the lower electrode 102 and the upper electrode 104 can be set as a parameter of a specific value, and is configured as a control parameter as described above. A refrigerant flow path 103A connected to the refrigerant piping 114 is formed inside the support 103, and the refrigerant is circulated in the refrigerant flow path 103A through the refrigerant piping i14 to adjust the wafer W to a specific temperature. A gas flow path 103B is formed in the support body 103, the insulating material 102A, the lower electrode 102, and the electrostatic chuck 108, and the gas is introduced into the electrostatic chuck by the gas introduction mechanism 115 through the gas pipe n5A as a back gas at a specific pressure. The gap between 108 and the wafer W allows the He gas to increase the thermal conductivity between the electrostatic chuck 108 and the wafer W. In addition, i〗 6 series bellows cover. A gas introduction part 104A is formed on the shower head 104, and the gas introduction part 104A is connected to a process gas supply system 1 1 8 through a pipe 1 1 7. Process gas supply system: The 8 series has Ar gas supply sources 1 1 8 A, C 0 r gas supply sources 1 1 8 B, C 4 F 6 gas supply sources 1 1 8 C, and 0 2 gas supply sources 1 18D. These gas supply sources 1 18A, 1 18B, 118C, 118D pass through valves H8E, 118F, 118G, 118H and mass flow controllers 1 1 8 I, 1 1 8 J, 1 1 8 K, 1 1 8 L, respectively. The flow rate is set to supply individual gases to the shower head 104, and the inside thereof is adjusted to a mixed gas having a specific mixing ratio. Each gas flow is a parameter that can be controlled by individual mass flow control -16-(13) (13) 200404333 controller 1 1 81, 1 1 8J, 1 18K, 1 18L, as described above, It is constituted as a control parameter. Below the above-mentioned shower head 104, a large number of holes 104b are arranged comprehensively and evenly. Through these holes 104B, the shower head 104 supplies the mixed gas into the upper chamber 101A as a process gas. In addition, an exhaust pipe 1 0 1 C is connected to an exhaust hole in the lower part of the lower chamber 101B, and the exhaust processing chamber 101 is exhausted through an exhaust system 119 formed by a vacuum pump or the like connected to the exhaust pipe 101C to maintain a specific gas pressure. An APC valve 101D is installed in the exhaust pipe 1 0 1 C, and the opening degree is automatically adjusted according to the gas pressure in the processing chamber 101. This opening degree is a device status parameter that displays the status of the device and is a parameter that cannot be controlled. A detection window 1 2 1 is provided on the side wall of the processing chamber 101, and the detection window 1 2 1 is penetrated outside the side wall of the processing chamber 1 101 to detect the plasma light emission in the processing chamber 1 0 across multiple wavelengths. Beamsplitter (hereinafter referred to as "optical detector") 1 2 0. The state of the plasma is monitored based on optical data on a specific wavelength obtained by the optical detector 120, for example, detecting the end of plasma processing. This optical data constitutes an electric prize reflection parameter reflecting the state of the electric prize together with the electric data based on the plasma generated by the high-frequency power P. Next, a multivariate analysis means provided in the plasma processing apparatus 100 will be described with reference to the drawings. The plasma processing apparatus 100 includes, for example, a multivariate analysis means 200 as shown in FIG. This multivariate analysis means 2 0 0 is provided with: a multivariate analysis program memory means 20 1 that memorizes a multivariate analysis program; and intermittent sampling by a control parameter measuring device 22 丨, plasma counter-17- (14) (14 200404333 The control parameter signal sampling means 202, the plasma reflection parameter signal sampling means 203, and the device state parameter signal sampling means 204 are the control parameter signal sampling means 202 of the detection signal from the reflection parameter measuring device 222 and the device state parameter measuring device 223. In addition, it includes analysis results or models necessary to analyze a large number of plasma reflection parameters (electrical data and optical data), a plurality of device state parameters related to the device state, and a plurality of control parameters. Analytical data memory means 205; and an arithmetic means 206 for calculating control parameters, plasma reflection parameters and device state parameters according to the purpose through the model; and an arithmetic signal from the arithmetic means 206 for the control parameters and the majority of the Prediction, diagnosis and control of plasma reflection parameters and device state parameters. Prediction, diagnosis and control methods. In addition, the multi-variable analysis means 200 is connected to the control means 22, the alarm means 226, and the display means 224 for controlling the plasma processing apparatus 100 according to the control parameters. The processing device control means 225 continues or interrupts the processing of the wafer W based on a signal from the prediction, diagnosis, and control means 207, for example. As described below, the alarm 226 and the display means 224 are based on the signal from the prediction, diagnosis, and control means 207, and are based on the purpose of notifying any of the control parameters, most of the plasma reflection parameters, and device status parameters. In addition, the analysis data storage means 205 stores data on the above parameters or processing data (processing data used for multivariate analysis). In addition, the control parameter measuring device 22 1, the plasma reflection parameter measuring device 222, and the device state parameter measuring device 223 control the majority of the flow rate detector, optical measuring device, and high-frequency voltage Vpp measuring device, respectively. The parameter measuring device, most of the plasma measuring parameters reflect the parameter, and most of the -18- (15) (15) 200404333 device state parameter measuring devices are aggregated into one and displayed. Here, the principle of the present invention will be described. For example, a plasma processing apparatus 100A is considered as a reference processing apparatus when creating a new model, and a plasma processing apparatus 100B is considered as a processing apparatus other than the reference processing apparatus. There is a slight individual difference between the plasma processing apparatuses 100A and 100B due to manufacturing variations and the like. In addition, the above-mentioned sensors such as the electrical measuring device 107C and the optical detector 120 also have individual differences due to manufacturing errors, etc., so even if the same type of plasma processing device uses the same type, Sensor, can not get the same test data. Therefore, even if it is the same type of plasma processing device, each plasma processing device needs to create a multivariate analysis model, and a multivariate analysis model cannot be used as a multivariate analysis model of other plasma processing devices of the same type. Therefore, in this embodiment, for example, between the plasma processing apparatuses 100A & 100B, even if there are individual differences in manufacturing, or there are individual differences among the individual majority sensors, the plasma processing apparatus The source of the multivariate analytical model made by 100A is used in other plasma processing devices 100B. In this embodiment, one method of multivariate analysis is to use a partial least square method (hereinafter referred to as "PLS (Partial Least Squares) method") to create an individual multivariate analysis model for the plasma processing apparatuses 100A and 100B. Find the individual difference between the devices and make a model that absorbs the individual difference. Details of the PLS method are disclosed in, for example, JOURNAL OF CHEMOMETRICS, VOL. 2 (PP. 2 1 1 -228) (1 998). For example, each of the plasma processing devices 100A and 100B has a large number of control parameters (-) (16) 200404333 (setting data) as the objective variables, and a large number of plasmas reflecting electrical and optical data detection data) as explanatory variables. The objective variable is the rank X of the component and the regression formula (hereinafter, "single type") shown in the following (1) with the explanatory variable as the component (the first process). In the individual calculations of the plasma processing apparatuses 100A and 100B, according to the PLS method using one method of multivariate analysis, the explanatory variables and objective variables obtained in the experiments are used to calculate the module rows K and Kb, respectively. Memorized in analysis means 205. In addition, in the following models (1) and (2), the regression models of individual models are shown, where a is a plasma processing device and b is a plasma processing apparatus 100B.
Xa = KaYa··· (1) Xb = KbYb·.· (2) P L S法係於行列X、Y中，即使個別有多數 數及目的變數，只要具有個別之少數的實測値， 得行列X和行列Υ之關係式。而且，即使爲以 測値所獲得之關係式，其穩定性及可靠性高， 法之特徵。在實測成爲說明變數及目的變數之各 分配控制參數以檢測控制參數，以多數感測器分 漿反映參數。 在此情形下，在分配控制參數（高頻電力、 參數（含 作成使以 之行列Υ 稱爲「模 手段 206 在個別的 型之回歸 資料記憶 ，Ka、Kb I 100A 、 的說明變 便可以求 少數的實 此爲 PLS 資料時， 節檢測電 處理室內 -20- (17) (17)200404333 壓力、製程氣體流量等）之範圍窄時，如下述式（3 ) $ 示般，可以對於控制參數以線性形式予以近似之，在分gg 參數之範圍大時，如下述式（4 )所示般，對於控制參數 可以置入平方、立方及1次和2次交叉項之非線性形式予 以近似之。 此種控制參數在電漿處理裝置100A和電漿處理裝霞 1 0 0B係使用相同範圍、相同値之控制參數。在求得回歸 行列Ka、Kb時’可以與本案申請人在日本專利特願 2001-398608號說明書中所提案的PLS法相同的運算步驟 而求得。此處’省略該運算步驟之說明。電漿處理裝置 100A和電漿處理裝置100B之間的個體差及個別之感測器 間的個體差，成爲上述式（1 )、式（2 )之回歸行列 Ka、Kb之差而表現。 X= [X 1，x2，…，xn] ··· (3 ) X= [x 1，x2， ··· ，xn， (xl)2,（x2)2，…，（xn)2， (x 1 )3, (x2)3，…，（xn)3， x1x2 ? x1x 3,…，xn-1χη? (xl)2x2? (xl)2x3··· (xn-l)2xn] ... (4) 然後’在藉由p L S法以求得上述模型時，事先藉由 利用晶圓之訓練裝置（training set )的實驗，以測量多數 的說明變數和多數的目的變數。爲此’例如作爲訓練裝 -21 - (18) (18)200404333 置，準備18片之晶圓（ΤΗ-OX Si)。另外，ΤΗ-OXSi係 形成熱氧化膜之晶圓。在此情形下’利用實驗計畫法’有 效率設定控制參數（設定資料）’可以最小限度的實驗完 成。 在電漿處理裝置1 00A中，例如在以標準値爲中心而 於特定範圍中，每一個訓練晶圓分配成爲目的變數之控制 參數，以蝕刻處理訓練晶圓。然後，在蝕刻處理時，關於 各訓練晶圓各多數次測量製程氣體之各氣體的流量、處理 室內的壓力等之控制參數、電氣資料及光學資料等的電漿 反映參數，透過運算手段2 06算出這些控制參數、電漿反 映參數的平均値。然後，將控制參數的平均値當成設定資 料使用，將電漿反映參數當成檢測資料使用。 分配控制參數之範圍係假定在進行蝕刻處理時，控制 參數最大限度變動之範圍，在此假定的範圍內，分配控制 參數。在本實施形態中，將高頻電力、處理室內壓力、上 下兩電極102、104間的間隙尺寸及各製程氣體（Ar氣 體、CO氣體、C4F6氣體及02氣體）的流量當成控制參數 (設定資料）使用。各控制參數之標準値因鈾刻對象而不 同。在電漿處理裝置100B也以與電漿處理裝置1〇〇A相 同之要領’以同一控制參數（設定資料）進行實驗，獲得 控制參數（設定資料）及電漿反映參數（檢測資料）。 具體爲，以標準値爲中心而在下述表1所示之等級1 及等級2之範圍內，每一訓練晶圓分配控制參數予以設 定’進行各訓練晶圓之蝕刻處理。然後，在處理各訓練晶 -22- (19) 200404333 圓間，透過電氣量測器107C將基於電漿之高頻電壓（由 基本波至4倍波爲止）V、高頻電流（由基本波至4倍波 爲止）I、相位差4等之電氣資料當成檢測資料予以測量 的同時’透過光學檢測器1 2 〇例如將2 0 0〜9 5 0 n m之波長 範圍的發光光譜強度（光學資料）當成檢測資料予以測 量，將這些檢測資料（電氣資料及光學資料）當成電漿反 映參數使用。另外’同時利用個別之控制參數量測器2 2 1 以測量下述表1所示之各控制參數° (表1 ) 電力 壓力 間隙 Ar CO C4F6 02 W m T 〇 rr mm seem seem seem seem 等級1 1400 3 8 25 1 70 36 9.5 3.5 標準値 1500 40 27 200 50 10 4 等級2_ 1540 42 29 230 64 10.5 4.5 2.67% 5.0% 7.41% 15.00% 2 8.0 0 % 5.0 0 % 12.50% 然後，在處理訓練晶圓時，將上述各控制參數設定爲 熱氧化膜之標準値，以標準値事先處理5片之僞晶圓’以 謀求電漿處理裝置l〇〇A、l〇〇B之穩定化。接著，在電漿 處理裝置100 A、10 0B中，進行18片之訓練晶圓的蝕刻 處理。此時，在電漿處理裝置100A中，如下述表2所示 般’在上述等級1及上述等級2之範圍內，每一訓練晶圓 分配上述各控制參數，即製程氣體（Ar、CO、C4F6、 (20) (20)200404333 〇2)流釐、處理室內之壓力、高頻電力，以處理各訓練晶 圓。 接著，關於各訓練晶圓，由個別之量測器獲得多數的 電氣資料及多數的光學資料。這些例如當作實測値而記憶 在解析資料記憶手段205。然後，在運算手段206中，算 出多數之控制參數個別的實測値之平均値、多數的電漿反 映參數（電氣資料、光學資料）個別之實測値的平均値， 將這些平均値當成目的變數及說明變數，記憶在解析資料 記憶手段2 0 5中。接著，在運算手段2 0 6中，利用P L S 法，依據這些運算資料，求得上述（1 )之模型的回歸行 列Ka(第1工程）。 另外，在電漿處理裝置100B中也與電漿處理裝置 100A相同’如下述表2所示般’分配控制參數’算出各 參數的實測値之平均値’將适些平均値當成目的變數及說 明變數使用，求得上述（2)之模型的回歸行列Kb(第1 工程）。另外，在下述表2中’ L 1〜L 8係表示訓練晶圓之 鉍碼。 (21)200404333 (表2 ) No. 壓力 Ar CO C 4 F 6 〇2 間隙 電力 [m T o rr ] [seem] [seem] [seem] [seem] [mm] [W] L 1 42 1 70 64 10 4.5 25 1500 L2 38 200 36 9.5 4.5 29 1500 L3 40 230 64 9.5 3.5 27 1500 L4 42 1 70 50 9.5 4.5 27 1540 L5 3 8 1 70 36 9.5 3.5 25 1460 V6· - 我 3 8 200 50 丫 27 15 00 L7 3 8 230 50 10 3.5 25 1540 L8 36 230 64 10.5 4.5 29 1540 L9 42 200 64 10 3.5 29 1460 L 1 0 40 1 70 50 10.5 3.5 29 1500 L 1 1 40 200 54 9.5 4 25 1540 Li 2 '42 1 200 56 10.5 3.5 2 7 1540 L 1 3 42 230 36 10.5 4 25 1500 L 1 4 40 230 36 10 1.5 27 1460 L 1 5 40 200 50 10.5 4.5 25 1460 L 1 6 42 230 50 9.5 3.5 29 1460 L 1 7 40 170 36 10 3.5 29 1540 1 δ j δ 1 / u D 4 l υ . d ^ . 3 z / 1 0 u 求得回歸行列 Ka、Kb後，利用電漿處理裝置 1 00A，在下述表3所示之新的製程條件下，如下述表3 -25- (22) 200404333 所示般，由標準値分配製程氣體流量等之控制參數，處理 20片之測試晶圓（ΤΗ-OX Si )，藉由個別之感測器檢測 此時的電漿反映參數及裝置狀態參數。此時，如下述表3 所示般，將多數的控制參數設定爲製程條件的標準値，使 電漿處理裝置運轉，將5片之裸矽晶圓當成僞晶圓流通於 處理室101內，使電漿處理裝置穩定化。 -26- (23)200404333 (表3 ) 電力 壓力 間隙 Ar CO C 4 F 6 O 2 NO. W m To rr mm seem seem seem seem Bare Si 1 2000 100 35 300 50 10 8 Bare Si 2 2000 100 35 300 50 10 8 Bare Si 3 2000 100 35 300 50 10 8 Bare Si 4 2000 100 35 300 50 10 8 Bare Si 5 2000 100 35 300 50 10 8 TH-OX Si 6 2000 100 35 300 50 10 8 TH-OX Si 7 1980 100 35 300 50 10 8 TH-OX Si 8 1900 100 35 300 50 10 8 TH-OX Si 9 1980 100 35 280 50 10 8 TH-OX Si 10 2000 95 35 300 50 10 8 TH-OX Si 11 2000 100 3 3 300 50 10 8 TH-OX Si 12 2000 100 37 300 50 10 8 TH-OX Si 13 2000 100 35 270 50 10 8 TH-OX Si 14 2000 98 35 300 50 10 8 TH-OX Si 15 2000 100 35 300 50 10 8 TH-OX Si 16 2000 100 35 300 70 10 8 TH-OX Si 17 2000 100 35 300 50 8 8 TH-OX Si 18 2000 100 35 300 50 1 2 8 TH-OX Si 19 1900 95 35 300 50 10 6 TH-OX Si 20 1980 102 35 300 50 10 10 TH-OX Si 2 1 1900 98 33 300 50 10 10 TH-OX Si 22 1980 98 33 300 50 10 8 TH-OX Si 23 1900 100 35 270 50 10 8 TH-OX Si 24 1980 100 35 350 50 10 8 TH-OX Si 25 2000 100 35 300 50 10 8 -27- (24) (24)200404333 即在將處理室101內之上下電極102、104之間隙設 定爲3 5mm後，一開始電漿處理裝置的運轉時，支持體 103透過滾珠導螺桿機構112下降至處理室1〇1之下室 1 0 1 B的同時，由閘門閥1 0 6開放之出入口搬入僞晶圓， 載置在下部電極102上。晶圓W搬入後，閘門閥106關 閉的同時，排氣系統1 1 9動作，將處理室1 〇 1內維持在特 定的真空度。藉由此排氣，APC閥門101D之開度依據排 氣量而自動進行調整。此時，由氣體導入機構115將He 氣體當成背景氣體供給，提高晶圓W和下部電極1 02，具 體爲靜電夾頭1 〇 8和晶圓W間的熱傳導性，以提高晶圓 W的冷卻效率。 然後’由製程氣體供給系統118分別以300sccm、 50sccm、lOsccm以及8sccm之流量供給Ar氣體、CO氣 體、（：4116氣體以及02氣體。此時，將處理室101內之製 程氣體的壓力設定爲lOOrnTorr之故，APC閥門101D之 開度便依據製程氣體供給量和排氣量而自動調整。在此狀 態下，如由高頻電源107施加2000W之高頻電力時，與 偶極子環型磁鐵105之作用相輔，發生磁控管放電，而產 生製程氣體的電漿。開始爲裸矽晶圓之故，不進行蝕刻處 理。在特定時間（例如，1分鐘）處理裸晶源後，以與搬 入時相反的操作，將處理後之晶圓W由處理室101內搬 出，以相同條件處理至後述的第5片之僞晶圓爲止。 藉由僞晶圓之處理，電紫處理裝置穩定後，處理測試 晶圓。關於最初的測試晶圓（即爲第6片晶圓），將控制 -28- (25) (25)200404333 參數維持爲標準値之原樣，進行蝕刻處理。在進行此處理 之間，透過電氣量測器1 0 7 C以及光學檢測器1 2 〇將電氣 資料以及光學資料分別當成檢測資料多數次予以測量，以 未圖示出之記憶手段記憶這些量測値。然後，依據這些量 測値，利用運算手段206算出平均値。 在處理第2片之測試晶圓時，將高頻電力由〗5 〇 〇 w 改變爲設定値1 9 8 0W，其他的控制參數係以上述的標準値 來進行蝕刻處理。在其間，與最初的測試晶圓相同，將電 氣資料以及光學資料當成檢測資料予以量測後，算出個別 之平均値。 在處理第3片以後的測試晶圓時，每次都如表3所示 般分配設定各控制參數，蝕刻處理各測試晶圓，關於各測 試晶圓，將電漿反映參數（電氣資料、光學資料）當成檢 測資料予以量測，算出個別之平均値。 與上述（1 )之模型相同，由此種控制參數的平均値 之行列Xa’和電漿反映參數的平均値之行列 Ya’作成下述 (5 )所示之新的模型（第2工程）。Xa = KaYa ... (1) Xb = KbYb ... (2) The PLS method is in the ranks X, Y. Even if there are individual majority numbers and objective variables, as long as there are individual measured values of the minority, the ranks X and Relations of ranks. In addition, even if it is a relational expression obtained by measuring, its stability and reliability are high, and the characteristics of the method. In the actual measurement, each of the explanatory variables and the target variable is assigned a control parameter to detect the control parameter, and the parameters are reflected by the majority of sensors. In this case, when the distribution of control parameters (high-frequency power, parameters (including preparations) is called "modular means 206 in the regression data memory of individual models, the description of Ka, Kb I 100A, and When a few of them are PLS data, the range of -20- (17) (17) 200404333 pressure, process gas flow, etc. in the energy-saving test processing room is narrow, as shown in the following formula (3) It is approximated in a linear form. When the range of the gg parameter is large, as shown in the following formula (4), the nonlinear form of the control parameter that can be placed in the square, cubic, and first and second cross terms is approximated. This kind of control parameter uses the same range and the same control parameter in the plasma processing device 100A and the plasma processing equipment Xia 100B. When the regression ranks Ka and Kb are obtained, it can be compared with the applicant in this case in Japanese patent May be obtained from the same calculation steps of the PLS method proposed in the 2001-398608 specification. The description of the calculation steps is omitted here. Individual differences and individual differences between the plasma processing apparatus 100A and the plasma processing apparatus 100B The individual difference between the sensors is expressed as the difference between the regression rows Ka and Kb in the above formulas (1) and (2). X = [X 1, x2, ..., xn] (3) X = [x 1, x2, ..., xn, (xl) 2, (x2) 2, ..., (xn) 2, (x1) 3, (x2) 3, ..., (xn) 3, x1x2? x1x 3, ..., xn-1χη? (xl) 2x2? (xl) 2x3 ... (xn-l) 2xn] ... (4) Then, when the above model is obtained by the p LS method, Experiments using a training set of wafers to measure most explanatory variables and most target variables. For this purpose, for example, as a training device -21-(18) (18) 200404333, prepare 18 pieces of Wafer (T-OX Si). In addition, T-OXSi is a wafer on which a thermal oxide film is formed. In this case, 'use the experimental planning method' to efficiently set control parameters (setting data) 'can be completed in a minimum of experiments. In the plasma processing apparatus 100A, for example, in a specific range centered on a standard radon, each training wafer is assigned a control parameter as a target variable, and the training wafer is etched. Then, during the etching process, About each training The wafers each measure the flow rate of each gas in the process gas, the control parameters such as the pressure in the processing chamber, and the plasma reflection parameters such as electrical data and optical data. These control parameters and plasma reflection parameters are calculated by computing means 20 06. Average 値. Then, the average 値 of the control parameters is used as the setting data, and the plasma reflection parameters are used as the detection data. The range of the distribution of the control parameters is assumed to be the range in which the control parameters are maximally changed during the etching process. Within the range assumed, the control parameters are allocated. In this embodiment, the high-frequency power, the pressure in the processing chamber, the size of the gap between the upper and lower electrodes 102 and 104, and the flow rate of each process gas (Ar gas, CO gas, C4F6 gas, and 02 gas) are used as control parameters (setting data) )use. The standard plutonium of each control parameter differs depending on the engraved object. In the plasma processing apparatus 100B, the same method as that of the plasma processing apparatus 100A is used to conduct experiments with the same control parameters (setting data) to obtain the control parameters (setting data) and the plasma reflection parameters (testing data). Specifically, with the standard 値 as the center and within the range of level 1 and level 2 shown in Table 1 below, each training wafer allocation control parameter is set 'to etch the training wafers. Then, during the processing of each training crystal-22- (19) 200404333, the high frequency voltage (from the fundamental wave to 4 times the wave) based on the plasma and the high frequency current (from the fundamental wave) Up to 4 times the wave) I, phase difference 4 and other electrical data are measured as detection data and transmitted through the optical detector 1 2 〇 For example, the light emission spectral intensity in the wavelength range of 200 to 9 50 nm (optical data ) Measure as test data, and use these test data (electrical data and optical data) as plasma reflection parameters. In addition, at the same time, an individual control parameter measuring device 2 2 1 is used to measure each control parameter shown in Table 1 below (Table 1) Power pressure gap Ar CO C4F6 02 W m T 〇rr mm seem seem seem seem level 1 1400 3 8 25 1 70 36 9.5 3.5 Standard: 1500 40 27 200 50 10 4 Level 2_ 1540 42 29 230 64 10.5 4.5 2.67% 5.0% 7.41% 15.00% 2 8.0 0% 5.0 0% 12.50% When the circle is round, the above-mentioned control parameters are set as the standard of the thermal oxide film, and 5 pseudo wafers are processed in advance according to the standard to stabilize the plasma processing apparatuses 100A and 100B. Next, in the plasma processing apparatuses 100 A and 100B, etching processing of 18 training wafers is performed. At this time, in the plasma processing apparatus 100A, as shown in Table 2 below, within the range of the above-mentioned level 1 and the above-mentioned level 2, each training wafer is assigned the above-mentioned control parameters, that is, the process gas (Ar, CO, C4F6, (20) (20) 200404333 〇2), pressure in the processing room, high-frequency power to process each training wafer. Next, for each training wafer, most of the electrical data and most of the optical data are obtained by individual measuring devices. These are stored in the analysis data storage means 205 as measured data, for example. Then, in the calculation means 206, the average value of the individual measured values of the majority of the control parameters and the average value of the individual measured values of the majority of the plasma reflection parameters (electrical data, optical data) are calculated, and these average values are used as the target variables and The explanatory variables are memorized in the analytical data memory means 205. Next, in the calculation means 206, the regression line Ka of the model (1) described above is obtained using the PLS method based on these calculation data (first process). In addition, the plasma processing apparatus 100B is also the same as the plasma processing apparatus 100A. 'Assign control parameters' as shown in Table 2 below. Calculate the average of the actual measured values of each parameter. The average values will be used as the target variables and descriptions. Use the variables to find the regression rank Kb of the model (2) (1st process). In the following Table 2, 'L 1 to L 8 are bismuth codes of the training wafer. (21) 200404333 (Table 2) No. Pressure Ar CO C 4 F 6 〇 2 Clearance power [m T o rr] [seem] [seem] [seem] [seem] [mm] [W] L 1 42 1 70 64 10 4.5 25 1500 L2 38 200 36 9.5 4.5 29 1500 L3 40 230 64 9.5 3.5 27 1500 L4 42 1 70 50 9.5 4.5 27 1540 L5 3 8 1 70 36 9.5 3.5 25 1460 V6 ·-I 3 8 200 50 Ah 27 15 00 L7 3 8 230 50 10 3.5 25 1540 L8 36 230 64 10.5 4.5 29 1540 L9 42 200 64 10 3.5 29 1460 L 1 0 40 1 70 50 10.5 3.5 29 1500 L 1 1 40 200 54 9.5 4 25 1540 Li 2 '42 1 200 56 10.5 3.5 2 7 1540 L 1 3 42 230 36 10.5 4 25 1500 L 1 4 40 230 36 10 1.5 27 1460 L 1 5 40 200 50 10.5 4.5 25 1460 L 1 6 42 230 50 9.5 3.5 29 1460 L 1 7 40 170 36 10 3.5 29 1540 1 δ j δ 1 / u D 4 l υ. D ^. 3 z / 1 0 u After obtaining the regression rows Ka and Kb, use a plasma processing device 100A. Under the new process conditions shown in Table 3, as shown in Table 3 -25- (22) 200404333, control parameters such as process gas flow are allocated by standard 値, and 20 test wafers (TΗ-OX Si) are processed. ), By individual sensor detection At this time, the plasma parameters and reflect device status parameters. At this time, as shown in Table 3 below, most control parameters are set to the standard of process conditions, the plasma processing device is operated, and five bare silicon wafers are circulated in the processing chamber 101 as pseudo wafers. Stabilizing the plasma processing apparatus. -26- (23) 200404333 (Table 3) Power pressure gap Ar CO C 4 F 6 O 2 NO. W m To rr mm seem seem seem seem Bare Si 1 2000 100 35 300 50 10 8 Bare Si 2 2000 100 35 300 50 10 8 Bare Si 3 2000 100 35 300 50 10 8 Bare Si 4 2000 100 35 300 50 10 8 Bare Si 5 2000 100 35 300 50 10 8 TH-OX Si 6 2000 100 35 300 50 10 8 TH-OX Si 7 1980 100 35 300 50 10 8 TH-OX Si 8 1900 100 35 300 50 10 8 TH-OX Si 9 1980 100 35 280 50 10 8 TH-OX Si 10 2000 95 35 300 50 10 8 TH-OX Si 11 2000 100 3 3 300 50 10 8 TH-OX Si 12 2000 100 37 300 50 10 8 TH-OX Si 13 2000 100 35 270 50 10 8 TH-OX Si 14 2000 98 35 300 50 10 8 TH-OX Si 15 2000 100 35 300 50 10 8 TH-OX Si 16 2000 100 35 300 70 10 8 TH-OX Si 17 2000 100 35 300 50 8 8 TH-OX Si 18 2000 100 35 300 50 1 2 8 TH-OX Si 19 1900 95 35 300 50 10 6 TH-OX Si 20 1980 102 35 300 50 10 10 TH-OX Si 2 1 1900 98 33 300 50 10 10 TH-OX Si 22 1980 98 33 300 50 10 8 TH-OX Si 23 1900 100 35 270 50 10 8 TH-OX Si 24 1980 100 35 350 50 10 8 TH-OX Si 25 2000 1 00 35 300 50 10 8 -27- (24) (24) 200404333 After the gap between the upper and lower electrodes 102 and 104 in the processing chamber 101 is set to 35 mm, the support 103 starts when the plasma processing apparatus starts to operate. At the same time as falling through the ball screw mechanism 112 to the processing chamber 101 below the chamber 101 B, the gate wafer 106 opened the entrance and exit to carry the dummy wafer and placed it on the lower electrode 102. After the wafer W is carried in, the gate valve 106 is closed, and the exhaust system 119 is operated to maintain a specific vacuum degree in the processing chamber 101. With this exhaust, the opening degree of the APC valve 101D is automatically adjusted according to the exhaust volume. At this time, the He gas is supplied as the background gas by the gas introduction mechanism 115 to improve the thermal conductivity between the wafer W and the lower electrode 102, specifically, the electrostatic chuck 108 and the wafer W, so as to improve the cooling of the wafer W effectiveness. Then, the process gas supply system 118 supplies Ar gas, CO gas, (4116 gas, and 02 gas) at a flow rate of 300 sccm, 50 sccm, 10 sccm, and 8 sccm, respectively. At this time, the pressure of the process gas in the processing chamber 101 is set to 100 rnTorr Therefore, the opening degree of the APC valve 101D is automatically adjusted according to the process gas supply and exhaust volume. In this state, when 2000W high-frequency power is applied from the high-frequency power source 107, it is in contact with the dipole ring magnet 105. Complementing the effect, a magnetron discharge occurs, and a plasma of the process gas is generated. It is not a silicon wafer, so no etching is performed. After processing the bare crystal source at a specific time (for example, 1 minute), In the opposite operation, the processed wafer W is carried out from the processing chamber 101 and processed under the same conditions up to the fifth pseudo-wafer described later. After the pseudo-wafer processing, the electric violet processing device is stabilized. Process the test wafer. Regarding the initial test wafer (ie, the sixth wafer), maintain the control -28- (25) (25) 200404333 parameters as standard, and perform the etching process. In the meanwhile, electrical data and optical data were measured as electrical and optical data by electrical measuring device 10 7 C and optical detector 12 2 respectively, and these measurements were memorized by unillustrated memory means. Then, according to For these measurement chirps, the average chirp was calculated using the calculation means 206. When processing the second test wafer, the high-frequency power was changed from 〖500w to the setting of 値 180 8, and the other control parameters were as described above. In the meantime, the same as the original test wafer, the electrical data and optical data are measured as the test data to calculate the individual average 値. When processing the third and subsequent test wafers Each time, as shown in Table 3, each control parameter is assigned and set, and each test wafer is etched. For each test wafer, the plasma reflection parameters (electrical data, optical data) are used as test data to measure and calculate the individual The average 模型 is the same as the model of (1) above. From the rank Xa 'of the average 値 of the control parameter and the Ya' of the plasma reflection parameter, the following (5 ) Shows the new model (Project 2).
Xa，= Ka，Ya，…（5) 接著，在將電漿處理裝置100B以與電漿處理裝置 1 0 0 A相同條件分配控制參數時，關於電漿處理裝置1 0 0 B 係可不用如電漿處理裝置100 A般進行實驗而援用電漿處 -29- (26) (26) 200404333 理裝置100A之上述（5 )所示的模型。即在電漿處理裝 置100B中，以與電漿處理裝置100A相同的條件分配控 制參數之故，下述（6)式在電漿處理裝置100B之目的變 數的行列Xb，成立。因此，電漿處理裝置100B之目的變 數在行列Xb’時，上述（2)之模型係成爲下述（7)式之 模型。 然後，由上述（1)式以及下述（5)式所示的電漿處 理裝置100 A之模型和上述（2)式以及下述（7)式所示 的電漿處理裝置1 0 0 B之模型的關係，可以獲得下述（8 ) 式所示之模型。即在電漿處理裝置100A之回歸行列Ka， 新的回歸行列Ka ’，和電漿處理裝置1 〇 〇 B之回歸行列 Kb ，新的回歸行列 Kb’之間’比例關係 (Kb，/Ka，=Kb/Ka)成立，所以 Kb’=Ka’Kb/Ka。如在下述 (7 )式之Kb，中適用此關係時，則可以獲得下述（8 )。Xa, = Ka, Ya, ... (5) Next, when the plasma processing apparatus 100B is assigned control parameters under the same conditions as the plasma processing apparatus 100A, the plasma processing apparatus 100B may be omitted. The plasma processing apparatus 100A performs experiments in general, and the plasma processing apparatus 29- (26) (26) 200404333 is used for the model shown in (5) above. That is, in the plasma processing apparatus 100B, the control parameters are allocated under the same conditions as the plasma processing apparatus 100A. Therefore, the following formula (6) holds in the rank Xb of the objective variable of the plasma processing apparatus 100B. Therefore, when the objective variable of the plasma processing apparatus 100B is in the rank Xb ', the model of the above (2) is a model of the following formula (7). Then, the model of the plasma processing apparatus 100 A shown by the above formula (1) and the following (5) and the plasma processing apparatus 100 by the formula (2) and the following (7) The relationship between the models can be obtained by the following formula (8). That is, the proportional relationship between the return rank Ka of the plasma processing apparatus 100A, the new return rank Ka ', and the return rank Kb of the plasma processing apparatus 1000B, the new return rank Kb' (Kb, / Ka, = Kb / Ka), so Kb '= Ka'Kb / Ka. When this relationship is applied to Kb in the following formula (7), the following (8) can be obtained.
Xb，=Xa， …（6)Xb, = Xa,… (6)
Xb，= Kb，Yb， …（7)Xb, = Kb, Yb, ... (7)
Xb’ = (Ka，Kb/Ka) Yb’ …（8) 因此，在以行列Xb’分配控制參數的新製程條件中， 關於電漿處理裝置100A，如求得模型（5 )，由事先所求 得之電漿處理裝置100A的模型（1 )和電漿處理裝置 1 00B之模型（2 )和上述模型（7 ) ’則如（8 )式所示 般，可以作成關於電漿處理裝置100B之新的模型（第3 -30 - (27) (27)200404333 工程）。 即藉由求得使關於以新的製程條件所檢測的電漿處理 裝置1 0 0 A的控制參數之平均値（設定資料）的行列 X a，，和電漿反映參數的平均値（檢測資料）之行列γ ^ 相關連的回歸行列Ka，，可以作成電漿處理裝置1 00B的 新的模型（8 )，藉由此新的模型（8 )可以評估電漿處理 裝置1 00B之裝置狀態。此係意味著如依據實驗作成關於 電漿處理裝置100A之上述模型（5 )，則關於電漿處理 裝置100B，即使不重新實驗，也可以將上述（8 )式當成 電漿處理裝置10 0B的新的模型予以作成。 如此作成的新模型（8 )也可以記憶在電漿處理裝置 10 0B的解析資料記憶手段205中。藉此，在藉由電漿處 理裝置100B之平常運轉以做晶圓處理時，在由多數的電 漿反映參數之個別平均値（檢測資料）預測計算多數的控 制參數値時，可以使用解析資料記憶手段205之新模型 (8)。 在此情形下，藉由預測·診斷·控制手段207比較所 預測的控制參數値（所預測的設定資料）和實際設定在電 漿處理裝置100B之設定資料的變動容許範圍，在判斷爲 異常時，例如藉由處理裝置控制手段225以停止電漿處理 裝置100B的同時，以顯示手段224、警報器226通知異 常。 如以上說明過的，在本實施形態中，具有：例如在電 漿處理裝置100A、100B中，分別基於第1設定資料（例 -31 - (28) (28)200404333 如，控制參數）而動作時，藉由多變量解析而每一電漿處 理裝置100A、100B求得由各電漿處理裝置100A、100B 之多數感測器所檢測的檢測資料（例如，電繁反映參數） 和第1設定資料的相關關係（（1 )式之Ka，、 （ 2 )式之Xb '= (Ka, Kb / Ka) Yb'… (8) Therefore, in the new process condition for assigning control parameters by rank Xb ', regarding the plasma processing device 100A, if the model (5) is obtained, The obtained model (1) of the plasma processing apparatus 100A, the model (2) of the plasma processing apparatus 100B, and the above model (7) 'are as shown in the formula (8), and the plasma processing apparatus 100B can be made. New model (Projects 3 -30-(27) (27) 200404333). That is, by obtaining the rank X a of the average 値 (setting data) of the control parameters of the plasma processing device 1 0 0 A detected under the new process conditions, and the average 値 (detection data) of the plasma reflection parameters A new regression model Ka related to the rank γ ^ can be used to create a new model (8) of the plasma processing apparatus 100B. From this new model (8), the device state of the plasma processing apparatus 100B can be evaluated. This means that if the above-mentioned model (5) of the plasma processing apparatus 100A is made based on experiments, the above-mentioned formula (8) can be regarded as the plasma processing apparatus 100B of the plasma processing apparatus 100B without re-experimentation. Create a new model. The new model (8) thus created can also be stored in the analysis data storage means 205 of the plasma processing apparatus 100B. With this, when wafer processing is performed by the normal operation of the plasma processing apparatus 100B, when the individual averages (inspection data) of most parameters reflected by the plasma are used to predict and calculate the majority of control parameters, analytical data can be used. New model of memory means 205 (8). In this case, the predicted control parameter 値 (predicted setting data) is compared with the allowable range of change of the setting data actually set in the plasma processing apparatus 100B by the prediction, diagnosis, and control means 207, and when it is judged that it is abnormal For example, when the plasma processing apparatus 100B is stopped by the processing apparatus control means 225, the abnormality is notified by the display means 224 and the alarm 226. As described above, in this embodiment, for example, the plasma processing apparatuses 100A and 100B operate based on the first setting data (eg, -31-(28) (28) 200404333, such as control parameters). In each case, each of the plasma processing apparatuses 100A and 100B obtains the detection data (for example, the electric propagation reflection parameter) and the first setting detected by the majority of the sensors of the respective plasma processing apparatuses 100A and 100B by multivariate analysis. Correlation of the data (Ka of (1), and (2) of
Kb )的第1工程；以及在各電漿處理裝置ιοοΑ、ιοοΒ中 之當成基準處理裝置的電漿處理裝置100A中，依據新的 第2設定資料（例如，分配控制參數之範圍與第丨設定資 料不同的新的設定資料）動作時，藉由多變量解析以求得 由電漿處理裝置1 0 〇 A的多數感測器所檢測的檢測資料和 第 2設定資料的相關關係（（5 )式之Ka’）的第2工 程；以及基於在第1工程所求得之電漿處理裝置100B的 相關關係Kb，和在第1工程所求得之電漿處理裝置100a 的相關關係 Ka，和在第2工程所求得知電漿處理裝置 100A的相關關係Ka’以求得基準處理裝置以外的其他處 理裝置之電漿處理裝置1 〇 〇 B的第2設定資料和檢測資料 的相關關係（（8 )式之Kb ’），作成依據如此求得之相 關關係Kb’，以評估電漿處理裝置100B的裝置狀態或者 預測處理結果之多變量解析模型（（8 )式）的第3工 程。 因此，關於藉由新製程條件的新的設定資料，在當成 基準之電漿處理裝置100A中，如對於晶圓進行電漿處理 之實驗而作成模型（5)時，援用電漿處理裝置100 A之 模型（5)，可以作成當成基準之處理裝置以外的處理裝 置，例如電漿處理裝置100B之新的模型（8 )。因此，關 -32- (29) 200404333 於電漿處理裝置100B，即使不爲了作成新的模型而 的設定資料進行實驗，也可以作成新模型（8 )。藉 可以大幅減輕關於電漿處理裝置100B之模型作成 擔。 另外，在本實施形態中，第3工程係依據對於在 工程所求得之電漿處理裝置100B的相關關係Kb的 設定資料和檢測資料的相關關係Kb，，和對於在第1 所求得之電漿處理裝置10 0A的相關關係Ka的第2 所求得之相關關係Ka，的比例關係，以求得電漿處理 1 00B之第2設定資料和檢測資料的相關關係Kb， 此’電漿處理裝置100B之相關關係Kb，可以不藉由 量解析而簡單算出。 另外，在本實施形態中，使可以控制電漿狀態之 的控制參數和反映電漿狀態的多數的電漿反映參數相 作成多變量解析模型。具體爲，利用電漿處理 1 〇 〇 A、1 〇 〇 b，分別以設定資料（控制參數等）爲目 數’同時以檢測資料（電漿反映參數等）爲說明變數 成多變量解析模型（1 ) 、（ 2 )。然後，在新的設定 中，關於電漿處理裝置100A而作成多變量解析 (5 )時，利用相關關係Kb，和設定資料xb，，算出電 理裝置1 00B之檢測資料（電漿反映參數、裝置狀態 等）’可以作成關於電漿處理裝置10 0B之藉由新的 資料的多變量解析模型（8 )。 另外，爲了利用PL S法以作成多變量解析模型 以新 此， 的負 第1 第2 工程 工程 裝置 。藉 多變 多數 關以 裝置 的變 ，作 資料 模型 漿處 參數 設定 ，即 -33- 3¾ (30) (30)200404333 使實驗數少，也可以高精度預測、評估上述各參數。另 外，藉由以電漿處理裝置1 00B之預測値爲主成分分析， 可以綜合地評估電漿處理裝置1 00B之運轉狀態。 另外，進行利用PLS法以求得設定資料和檢測資料 之相關關係的多變量解析，可以少的資料進行高精度的多 變量解析以作成多變量解析模型。 接著，一面參考圖面一面說明本發明之第2實施形 態。第3圖係顯示關於本實施形態之控制系統整體的槪略 構造方塊圖。此控制系統3 0 0係由藉由網路3 2 0以連接主 機裝置310和多數的電漿處理裝置100A.....100N而構 成。電漿處理裝置1〇〇 A.....1〇 ON係分別與第1圖所示 者同樣的構造故，省略其之詳細說明。另外，電漿處理裝 置100A、…、100N係分別具備如第2圖所示之多變量解 析手段200。另外，在本實施形態中，例如第2圖所示之 多變量解析手段200、處理裝置控制手段225、第3圖所 示之發送接收裝置1 5 0係擔負當成處理裝置的控制裝置之 任務。 主機裝置3 1 0係至少具備：進行各種運算的運算手段 3 1 2、記憶上述之PLS法等之多變量解析程式的多變量解 析程式記憶手段3 1 4、記憶解析結果或解析所必要的增要 的解析資料記憶手段3 1 6、透過上述網路320與各電漿處 理裝置100A.....100N進行資料之交換的發送接收手段 318。另外，上述主機裝置310例如可以半導體製造工廠 的主電腦構成，或以連接於此主電腦之個人電腦構成。 -34- (31) (31)200404333 電漿處理裝置100 A.....10 ON分別具備：進行在各 電漿處理裝置100 A.....10 ON和主機裝置310之間或考 各電漿處理裝置100A.....100N間的各種資料的發送接 收的發送接收裝置 150A、…、15〇Ν、輸入控制參數（設 定資料）等之各種資料用的輸入手段152A.....152N。 上述發送接收裝置150A.....15 0N係分別與第2圖所示 之多變量解析手段 200連接，可以與各電漿處理裝置 100A.....100N之多變量解析手段200進行資料之交 換。 上述網路3 20係可以雙向通訊以連接主機裝置310、 各電漿處理裝置100A.....100N之網路，典型上可舉網 際網路等之公眾線路網。另外，網路3 2 0在上述公眾線路 網之外，也可以爲 WAN(Wide Area Network:廣域網 路）、LAN(Local Area Network :區域網路）、IP-VPN(Internet Protocol-Virtual Private Network :網際網 路通訊協定-虛擬私人網路）等之封閉線路網。另外，對 於網路 3 2 0 之連接媒體，可以爲藉由 FDDI(Fiber Distributed Data Interface:光纖分配數據介面）等之光纖 電纜、藉由Ethernet之同軸電纜或者對絞電纜、或者藉由 IEEE8 02 lib等之無線等，不管有線無線，也可以爲衛星 網路等。 各電漿處理裝置1 00在以所期望的製程條件進行蝕刻 處理時，藉由將作成評估裝置狀態用之新模型所必要的資 料，由主機裝置310透過發送接收裝置150而發送給所期 -35- 00 (32) 200404333 望之電漿處理裝置1 〇 〇，可以減輕以該電漿處理裝置i 的多變量解析手段200作成模型時的負擔。而且，由藉 電發處理裝置1 00之實際的晶圓處理時，利用新的模型 評估裝置狀態’依據因應其結果而由預測.診斷·控制 段207所輸出的指示，藉由處理裝置控制手段22 5得以 制電漿處理裝置1 〇 〇。 接著’參考圖面說明此種控制系統3 〇 〇之處理。控 系統3 0 0之處理例如係如在第1實施形態說明過的，可 將在電漿處理裝置100A所作成的新模型援用於電漿處 裝置100B’以作成電漿處理裝置i〇〇B的新模型之例子 第4圖〜第6圖係顯示作成電漿處理裝置ι〇〇Β之 模型時的處理之動作流程。更詳細爲第4圖〜第6圖係 示以電漿處理裝置100A爲基準處理裝置，以電漿處理 置1GGB.....100N爲基準處理裝置以外的處理裝置時 主機裝置、基準處理裝置、基準處理裝置以外的處理裝 的動作流程。另外，在第4圖〜第6圖中，基準處理裝 以外的處理裝置係以電漿處理裝置1 〇 〇 B之處理爲代表 記載。在關於其他的電漿處理裝置100C.....100N要 成新模型時’也進行與電漿處理裝置100B同樣的動作^ 首先’如第 4圖所示般，求得各電漿處理裝 100 A.....1〇 ON之回歸行列Ka.....Kn。以下說明具 之處理。Kb) the first project; and the plasma processing device 100A serving as the reference processing device in each of the plasma processing devices ιοοΑ, ιοοΒ, based on the new second setting data (for example, the range of distribution control parameters and the first setting New setting data with different data) During operation, the correlation between the detection data detected by the majority of sensors of the plasma processing apparatus 100A and the second setting data is obtained by multivariate analysis ((5) Ka ′) of the second process; and the correlation Kb based on the plasma processing apparatus 100B obtained in the first process, and the correlation Ka of the plasma processing apparatus 100a obtained in the first process, and The correlation Ka 'of the plasma processing apparatus 100A was obtained in the second project to obtain the correlation between the second setting data and the testing data of the plasma processing apparatus 100B of the processing apparatus other than the reference processing apparatus ( (8) Eq. Kb '), based on the correlation Kb' obtained in this way, to evaluate the device state of the plasma processing apparatus 100B or the multivariate analysis model for predicting the processing result (Eq. (8)) . Therefore, regarding the new setting data based on the new process conditions, in the plasma processing device 100A used as a reference, for example, if a model of plasma processing is performed on a wafer (5), a plasma processing device 100 A is used. The model (5) can be made into a processing device other than the reference processing device, such as a new model (8) of the plasma processing device 100B. Therefore, Guan-32- (29) 200404333 can be used to create a new model in the plasma processing device 100B without experimenting with setting data for creating a new model (8). This can greatly reduce the burden of modeling the plasma processing apparatus 100B. In addition, in the present embodiment, the third process is based on the correlation Kb of the setting data and the detection data of the correlation Kb of the plasma processing apparatus 100B obtained in the process, and the correlation of Kb obtained by the first The correlation relationship Ka of the plasma processing device 100A is the second correlation ratio Ka, and the proportional relationship is obtained to obtain the correlation Kb of the second setting data and the detection data of the plasma processing 100B. This' plasma The correlation Kb of the processing device 100B can be easily calculated without using quantity analysis. In addition, in this embodiment, a multivariable analysis model is made by controlling parameters that can control the state of the plasma and plasma reflection parameters that reflect the majority of the state of the plasma. Specifically, the plasma treatment of 100A and 100b is performed using setting data (control parameters, etc.) as the number of meshes, and using detection data (plasma reflection parameters, etc.) as explanation variables into a multivariate analysis model ( 1), (2). Then, in the new setting, when a multivariate analysis (5) is performed on the plasma processing apparatus 100A, the correlation data Kb and the setting data xb are used to calculate the detection data (plasma reflection parameters, (Device status, etc.) 'A multivariate analysis model (8) about the plasma processing device 100B with new data can be created. In addition, in order to use the PLS method to create a multivariate analysis model, the new first and second engineering devices are negative. Based on the change of most devices, the parameters of the model are set as the data model, that is, -33- 3¾ (30) (30) 200404333 so that the number of experiments is small, and the above parameters can be predicted and evaluated with high accuracy. In addition, the analysis of the main components of the plasma processing apparatus 100B can be used to comprehensively evaluate the operation status of the plasma processing apparatus 100B. In addition, a multivariate analysis using the PLS method to obtain the correlation between the setting data and the detection data is performed, and a high-variable multivariate analysis can be performed with a small amount of data to create a multivariate analysis model. Next, a second embodiment of the present invention will be described with reference to the drawings. Fig. 3 is a block diagram showing a schematic configuration of the entire control system of this embodiment. This control system 300 is constituted by connecting the host device 310 and most of the plasma processing devices 100A, ..., 100N through a network 3 200. The plasma processing apparatus 100A ..... 10ON have the same structures as those shown in Fig. 1, and therefore detailed descriptions thereof are omitted. The plasma processing apparatuses 100A, ..., and 100N are each provided with a multivariate analysis means 200 as shown in FIG. Further, in this embodiment, for example, the multivariate analysis means 200 shown in Fig. 2, the processing device control means 225, and the transmission / reception device 150 shown in Fig. 3 are responsible for controlling the device as a processing device. The host device 3 1 0 is provided with at least: arithmetic means for performing various calculations 3 1 2. Multivariate analysis program memory means for memorizing multivariate analysis programs such as the above-mentioned PLS method 3 1 4. Memorization analysis results or additional necessary analysis The required analysis data storage means 3 1 6. The transmission and reception means 318 for exchanging data with each of the plasma processing apparatuses 100A ..... 100N through the above network 320. The host device 310 may be constituted by, for example, a host computer in a semiconductor manufacturing plant or a personal computer connected to the host computer. -34- (31) (31) 200404333 Plasma treatment device 100 A ..... 10 ON are provided respectively: between each plasma treatment device 100 A ..... 10 ON and host device 310 or test Sending and receiving devices 150A, ..., 150N, various input data for inputting control parameters (setting data), etc. among various plasma processing devices 100A, ..., 100N, and other data input means 152A ... ..152N. The above-mentioned sending and receiving devices 150A ..... 15 0N are connected to the multivariate analysis means 200 shown in FIG. 2 respectively, and can perform data with the multivariate analysis means 200 of each plasma processing device 100A ..... 100N. The exchange. The above-mentioned network 3 to 20 is a network that can communicate in both directions to connect the host device 310 and each of the plasma processing devices 100A ..... 100N, typically a public line network such as the Internet. In addition, the network 3 2 0 may be a wide area network (WAN), a local area network (LAN), or an Internet Protocol-Virtual Private Network (IP-VPN): Internet Protocol-Virtual Private Network) and other closed circuit networks. In addition, for the connection medium of the network 3 2 0, it can be an optical fiber cable such as FDDI (Fiber Distributed Data Interface), a coaxial cable or an Ethernet cable via Ethernet, or an IEEE8 02 lib Waiting wireless, etc., regardless of wired and wireless, can also be a satellite network. When each plasma processing device 100 performs an etching process under a desired process condition, the host device 310 transmits the data necessary for creating a new model for evaluating the state of the device to the desired device through the sending and receiving device 150- 35- 00 (32) 200404333 Wangzhi's plasma processing device 100 can reduce the burden when creating a model using the multivariate analysis means 200 of the plasma processing device i. In addition, when the actual wafer processing is performed by the power generation processing device 100, a new model is used to evaluate the state of the device, based on the instructions output by the prediction, diagnosis, and control section 207 based on the results, and the control means of the processing device is used. 22 5 was able to make a plasma processing apparatus 100. Next, the processing of such a control system 300 will be described with reference to the drawings. The processing of the control system 3 0 0 is, for example, as explained in the first embodiment, a new model created in the plasma processing apparatus 100A can be applied to the plasma processing apparatus 100B ′ to form a plasma processing apparatus i〇〇B Examples of new models Figures 4 to 6 show the operation flow of processing when a model of the plasma processing apparatus ι〇〇B is created. Figures 4 to 6 show the host device and the reference processing device when the plasma processing device 100A is used as the reference processing device, and the plasma processing device is set to 1GGB ..... 100N as the processing device other than the reference processing device. 2. The operation flow of processing devices other than the reference processing device. In addition, in FIGS. 4 to 6, the processing apparatuses other than the reference processing apparatus are representatively described by the plasma processing apparatus 1000B. When another plasma processing apparatus 100C ..... 100N is to be made into a new model, 'the same operation as that of the plasma processing apparatus 100B is also performed ^' First, as shown in FIG. 4, each plasma processing apparatus is obtained. The return ranks of 100 A ..... 10 ON are Ka ..... Kn. The following describes the processing.
基準處理裝置之電漿處理裝置100Α在求得回歸行 Ka用之設定資料（例如，控制參數）由輸入手段1 52 A 由 以 手 控 制 舉 理 〇 新 顯 裝 之 置 置 而 作 置 體 列 被 -36 - (33) (33)200404333 輸入而設定時，在步驟s 1 1 0中，依據此設定資料處理晶 圓W，取得檢測資料（例如，電漿反映參數），將這些設 定資料、檢測資料透過網路3 2 0而發送給主機裝置3 1 0。 另一方面，基準處理裝置以外的處理裝置之例如電漿 處理裝置1 00B在求得回歸行列Kb用之設定資料（例 如，控制參數）由輸入手段152 A被輸入而設定時，在步 驟S 5 1 0中，依據此設定資料處理晶圓W，取得檢測資料 (例如，電漿反映參數），將這些設定資料、檢測資料透 過網路320而發送給主機裝置310。 主機裝置 310在步驟 S210中，由電漿處理裝置 1 〇〇 A.....1 00N接收設定資料、檢測資料，記憶在解析 資料記憶手段3 1 6。接著，在步驟S 2 2 0中，藉由運算手 段3 1 2求的所接收之設定資料的每一晶圓之平均値，以這 些爲目的變數Xa.....Xn，記憶在解析資料記憶手段 3 1 6，同時，藉由運算手段3 1 2求得所接收之檢測資料的 每一晶圓之平均値，以這些爲說明變數 Ya.....Yn，記 億在解析資料記憶手段3 1 6。 接著’主機裝置310在步驟S230中，依據由多變量 解析程式記憶手段3 1 4之藉由PLS法的程式，與上述第1 實施形態相同地，藉由運算手段3 ；! 2由設定資料（目的變 數）Xa.....Xn、檢測資料（說明變數）Ya.....Yn求 得各電獎處理裝置1 00A、…、100N的回歸行列The plasma processing device 100A of the reference processing device is used to obtain the setting data (for example, control parameters) for the return line Ka. The input means 1 52 A is controlled by hand. The newly installed position is set as a body- 36-(33) (33) 200404333 When inputting and setting, in step s 1 10, the wafer W is processed according to the setting data to obtain inspection data (for example, plasma reflection parameters), and these setting data and inspection data Send to host device 3 1 0 via network 3 2 0. On the other hand, when a processing device other than the reference processing device, for example, a plasma processing device 100B, determines the setting data (for example, a control parameter) for returning to the rank Kb, it is set by inputting means 152 A, in step S5. In 10, the wafer W is processed according to the setting data to obtain detection data (for example, plasma reflection parameters), and these setting data and detection data are sent to the host device 310 through the network 320. In step S210, the host device 310 receives the setting data and the detection data from the plasma processing device 100A ..... 100N, and stores them in the analysis data storage means 3 16. Next, in step S 2 2 0, the average value of each wafer of the received setting data obtained by the arithmetic means 3 1 2 is calculated, and the variables Xa ..... Xn for these purposes are stored in the analysis data The memory means 3 1 6 and meanwhile, the average mean 手段 of each wafer of the received inspection data is obtained by the arithmetic means 3 1 2. These are used as explanation variables Ya ..... Yn. Means 3 1 6. Next, in step S230, the host device 310 uses the PLS method according to the program of the multivariate analysis program memory means 3 1 4 as in the first embodiment described above, and uses the arithmetic means 3; 2 by setting data ( Purpose variables) Xa ..... Xn, detection data (explanation variables) Ya ..... Yn find the regression ranks of each electric award processing device 1 00A, ..., 100N
Ka.....Kn，記憶在解析資料記憶手段3 1 ό。接著，在步 驟S240中’將這些設定資料Xa.....Xn、檢測資料 -37- (34) 200404333Ka ..... Kn, memorizing means of analyzing data in memory 3 1 ό. Next, in step S240, the setting data Xa ..... Xn, the detection data -37- (34) 200404333
Ya.....Yn、回歸行列Ka.....Kn透過網路3 2 0而 給各電漿處理裝置100Α、···、1()〇Ν。 基準處理裝置之電漿處理裝置100Α在步驟 中，由主機裝置3 1 0接收設定資料Xa、檢測資料Ya 歸行列Ka，當成如上述（1 )所示之模型予以記憶 外，基準處理裝置以外的處理裝置之例如電漿處理 100B在步驟S 5 20中，由主機裝置31〇接收設定 Xb、檢測資料Yb、回歸行列Kb，當成如上述（2 ) 之模型予以記憶。 接著，如第5圖所示般，作成基準處理裝置之電 理裝置100A的新模型。以下說明具體之處理。 電漿處理裝置100A在求得回歸行列Ka’用之新 定資料（例如，控制參數）由輸入手段1 52A被輸入 定時，在步驟S 1 3 0中，依據此設定資料處理晶圓w 得新的檢測資料（例如，電漿反映參數），將這些新 定資料、新的檢測資料透過網路3 20而發送給主機 3 1 0 0 主機裝置310在步驟S310中，由基準處理裝置 漿處理裝置1 00A接收新的設定資料、新的檢測資料 億在解析資料記憶手段3 1 6。接著，在步驟S 3 2 0中 由運算手段3 1 2求得接收之新的設定資料的每一晶圓 均値，將這些當成說明變數Xa’ .....Xn’，記憶在 資料記憶手段3 1 6，同時，藉由運算手段3 1 2求得接 新的檢測資料的每一晶圓的平均値，將這些當成目的 發送 S 1 20 、回 。另 裝置 資料 所示 漿處 的設 而設 ，取 的設 裝置 之電 ，記 ，藉 之平 解析 收之 變數 -38- (35) (35)200404333Ya ..... Yn and returning ranks Ka ..... Kn are given to each of the plasma processing apparatuses 100A, ..., 1 () ON via the network 3 2 0. In the step, the plasma processing device 100A of the reference processing device receives the setting data Xa and the detection data Ya from the host device 3 10 and returns to the rank Ka. It is memorized as a model as shown in the above (1). For example, the plasma processing 100B of the processing device receives the setting Xb, the detection data Yb, and the regression rank Kb by the host device 31 in step S520, and stores it as the model (2) above. Next, as shown in Fig. 5, a new model of the electrical device 100A of the reference processing device is created. The specific processing will be described below. When the plasma processing device 100A obtains new set data (for example, control parameters) for returning to the rank Ka ', the input means 152A is inputted at a timing, and in step S 1 30, the wafer w is newly processed according to the set data. The new set data and new test data are sent to the host 3 20 via the network 3 20 and the host device 310 in step S310. The reference processing device pulp processing device 1 00A receives new setting data, new detection data, and analyzes data storage methods 3 1 6. Next, in step S 3 2 0, each of the wafers that have received the new setting data is calculated by the calculation means 3 1 2, and these are regarded as explanatory variables Xa '..... Xn' and stored in the data memory. Means 3 1 6, meanwhile, the average mean value of each wafer receiving the new inspection data is calculated by means of calculation means 3 1 2, and these are sent as S 1 20 and returned as the purpose. In addition, the equipment is shown in the equipment information. Take the electricity of the equipment, write down, and then analyze the received variables. -38- (35) (35) 200404333
Ya’ .....Yn’，記憶在解析資料記憶手段316。 接著，主機裝置310在步驟S330中，依據多變量解 析程式記憶手段314之藉由PLS法的程式，與上述第1 實施形態相同地，由新的設定資料（目的變數）Xa’ '新 的設定資料（說明變數）Ya’藉由運算手段312以求得電 漿處理裝置1 〇〇A的回歸行列Ka’，記憶在解析資料記憶 手段 316。接著，在步驟 S340中，將這些新的設定資料 Xa’、新的檢測資料Ya’、新的回歸行歹ϋ Ka’透過網路320 而發送給電漿處理裝置100A。 基準處理裝置之電漿處理裝置100A在步驟 S140 中，由主機裝置310接收設定資料Xa’、檢測資料Ya’、 回歸行列Ka’，當成新模型予以記憶。 接著，如第6圖所示般，求得基準處理裝置以外的處 理裝置之例如電漿處理裝置100B的模型。基準處理裝置 以外的處理裝置之新模型係依據基準處理裝置的新模型而 求得之故，在基準處理裝置以外的處理裝置中，不需要重 新進行對於晶圓之電漿處理。以下說明具體之處理。 電漿處理裝置100B在步驟S 5 3 0中，在求得回歸行 列Kb ’用之設定資料（與求得回歸行列Kb ’用之設定資料 相同的設定資料）由輸入手段152B被輸入時，將此設定 資料透過網路320而發送給主機裝置310。 主機裝置310在步驟S4 10中，由基準處理裝置以外 的處理裝置之電漿處理裝置100B接收新的設定資料，記 憶在解析資料記憶手段3 1 6，藉由運算手段3 1 2求得接收 -39- (36) (36)200404333 的新設定資料的每一晶圓之平均値，將這些當成設定資料 (說明變數）X b ’ .....Xn ’，記憶在解析資料記憶手段 3 16° 接者’主機裝置310在步驟S420中，由基準處理裝 置以外的處理裝置之回歸行列（Kb.....Kn )、新的回 歸行列（Kb’ .....Kn’） ’和基準處理裝置的回歸行列 (Ka )、基準處理裝置的新的回歸行列（Ka，）之比例關 係（例如（Kb’/Ka’=Kb/Ka ))由運算手段312分別求得 新的回歸行列（K b ’ .....Κ η ’）。例如，電漿處理裝置 100Β 之（7 )式所示的新的回歸行列 Kb’係藉由 Kb’ = Ka’Kb/Ka所求得。藉此，在求得基準處理裝置以外 的處理裝置之新的回歸行列時，不需要重新進行PLS法 等之多變量解析處理，可以簡單求得。 接著，主機裝置310在步驟S430中，依據上述（7) 式所示之模型，由新的設定資料（Xb’ Xn’）、新 的回歸行列（Kb’ .....Kn’）算出新的檢測資料 (Yb ’ .....Yn ’），記憶在解析資料記憶手段3 1 6，將 這些新的設定資料（Xb’ .....Xn’）、新的回歸行列 (Kb， .....Kn’）算出新的檢測資料（Yb’ .....Υη’） 透過網路 3 20 分別發送給對應之電漿處理裝置 1 00B.....1 00N ° 例如在電漿處理裝置iOOB中，在步驟S540中，由 主機裝置310接收新的設定資料（Xb’ .....Xu’）、新 的回歸行列（Kb’ .....Kn’）算出新的檢測資料 -40- (37) 200404333 (Yb’ .....Yn’），當成如上述（8 )式所示之新 予以記憶。如此，在基準裝置以外的處理裝置中，得 成適合個別之處理裝置的新模型。 接著，參考圖面說明依據如此獲得之新模型，評 置狀態時的控制系統之處理。第7圖係顯示依據各電 理裝置所分別作成的新模型以評估裝置狀態時的主機 之動作流程和各電漿處理裝置的動作流程。 首先，在某電漿處理裝置100中，在步驟S610 對於設定資料的標準條件之容許變動範圍一被輸入時 記憶此容許變動範圍。此容許變動範圍係使用於判定 狀態爲正常或異常用之臨界値，例如，設爲對於分配 新模型用之設定資料，例如控制參數時的各控制參數 準値的最大値和最小値之範圍。 接著，上述電漿處理裝置1〇〇在步驟S 620中， 處理晶圓用的設定資料（標準條件，例如表1所示之 値）由輸入手段152被輸入時，依據此設定資料，電 理晶圓w ’取得每一晶圓所量測之設定資料和檢測資 將這些設定資料、檢測資料透過網路3 20發送給主機 3 10° 主機裝置310在步驟S710中，由上述電漿處理 1 〇 0每一晶圓的接收設定資料和檢測資料而記憶、在解 料記億手段3 1 6。然後，求得個別的平均値，當成設 料（目的變數）X’、檢測資料（說明變數）γ，，記億 析資料記憶手段3 1 6。接著，主機裝置3 1 〇在步驟 模型 以作 估裝 漿處 裝置 中， ，便 裝置 作成 的標 實際 標準 漿處 料， 裝置 裝置 析資 定資 在解 S720 -41 - (38) (38)200404333 中，將設定資料X ’、檢測資料Y ’發送給上述電漿處理裝 置 1 0 0。 電漿處理裝置100在步驟 S 63 0中，接收設定資料 X’、檢測資料Υ’，將這些當成實際的設定資料Xobs'、實 際的檢測資料Yobs’而記憶在解析資料記憶手段205。接 著，在步驟S640中，在上述（8)式所示之新模型適用實 際的檢測資料Yobs’，算出預測設定資料Xpre’，記憶在 解析資料記憶手段2 05。 接著，上述電漿處理裝置100在步驟S650中，藉由 預測設定資料Xpre’對於實際的設定資料Xobs’是否在容 許變動範圍內，以判定爲正常或異常。例如，預測設定資 料 Xpre’對於實際的設定資料Xobs’如在容許變動範圍 內，則判斷爲正常，如超過容許變動範圍，則判斷爲異 常。在判斷爲異常時，在步驟S660中，例如藉由處理裝 箧控制手段225，停止上述電漿處理裝置100，同時，以 顯示手段224、警報器226通知異常。 如此，主機裝置310依據來自各電漿處理裝置的資料 以求得平均値，而進行多變量解析處理，所以可以大幅減 輕各電漿處理裝置的計算處理負擔。另外，在各電漿處理 裝置中，不需要暫時記憶在進行電漿處理時所獲得的大量 設定資料或檢測資料等，也不需要多變量解析程式，所以 +需要爲此之記憶手段。藉此，可以使各電漿處理裝置之 擒造變得簡單，另外可以抑制製造成本。 另外，在第2實施形態中，雖就以各電漿處理裝置側 -42 - (39) 200404333 的新模型來判斷裝置狀態時做說明，但是並不一定 此，新模型也記憶在主機裝置310之故，也可在主 3 10側判定各電漿處理裝置1〇〇A.....100N之 態。在此情形時，於判定爲異常時，可將異常判定 送給各電漿處理裝置1〇〇A.....10ON。各電漿處 100 A.....1 0 ON可以依據異常判定資訊，例如藉 裝置控制手段225以停止處理裝置，藉由顯示手段 警報器226以通知異常。如依據此，可以主機裝置 中監視各電漿處理裝置之裝置狀態。 以上，雖一面參考所附圖面一面說明關於本發 適的實施形態，但是不用說，本發明並不限定於 子。只要是該行業者，在申請專利範圍所記載的範 很淸楚可以想到各種之變更例或者修正例，關於那 也應理解係屬於本發明之技術範圍。 例如，作爲上述第1及第2實施形態的設定資 如利用第2實施形態的新模型以判定裝置狀態時般 漿處理晶圓時，可以使用藉由控制參數量測器2 2 1 之設定資料，另外，也可以使用由輸入手段1 5 2所 設定資料本身。在此情形下，設定資料之全部都可 制參數量測器2 2 1測量時，雖可以使用藉由控制參 器22 1所測量的設定資料，但是在設定資料中，含 藉由控制參數量測器2 2 1所測量者時，以利用所輸 定資料本身爲有效。 另外，在上述實施形態的多變量解析中，雖然 限制於 機裝置 裝置狀 資訊發 理裝置 由處理 224、 310集 明之合 此種例 疇中， 些當然 料，係 ，在電 所測量 輸入的 藉由控 數量測 有無法 入的設 不使用 -43- (40) (40)200404333 裝置狀態參數，但是也可以將裝置狀態參數當成目的變數 或者說明變數使用。另外，在上述實施形態中，於構築模 型時，作爲目的變數之控制參數，雖使用製程氣體流量、 電極間的間隙及處理室內的壓力，但是只要是可以控制的 參數，並不限定於這些參數。 另外，作爲裝置狀態參數，雖利用可變電容器電容、 高頻電壓、APC開度，但是只要是顯示裝置狀態參數之可 測量的參數，並不限定於這些。另外，作爲反映電漿狀態 之電漿反映參數，雖使用基於電漿之電氣資料及光學資 料，但是只要是反映電漿狀態之參數，並不限定於這些。 另外，作爲電氣資料，雖使用基本波及高次諧波（至4倍 波爲止）之高頻電壓、高頻電流，但是並不限定於這些。 另外，也可將來自組裝在電漿處理裝置內的測量晶圓 完成結果之手段（例如，掃描量測儀）的輸出資料當成檢 測資料使用。具體爲作爲檢測資料可以使用形成在晶圓上 之膜的膜厚、蝕刻晶圓上的被處理膜時的蝕刻量或其面內 均勻性等之特徵値。另外，在本實施形態中，雖係每一晶 圓求得電漿反映參數之資料的平均値，使用此平均値而每 一晶圓預測控制參數及裝置狀態參數，但是也可以利用一 片之晶圓處理中的即時的電漿反映參數，即時地預測控制 參數及裝置狀態參數。 另外，在上述實施形態中，雖利用有磁場平行平板型 電漿處理裝置，但是只要是具有控制參數和電漿反映參數 及/或者裝置狀態參數之裝置，都可以適用本發明。 -44- (41) (41)200404333 如以上詳細說明般，如依據本發明，即時每一處理裝 置其製程特性或者製程條件有差異，如就一個處理裝置作 成模型，可將該模型援用於同種類的其他處理裝置，不需 要每一處理裝置重新作成模型，能夠提供可以減輕模型作 成之負擔的處理裝置之多變量解析模型作成方法及處理裝 置用之多變量解析方法。 產業上之利用可能性 本發明例如可以適用於電漿處理裝置等之處理裝置之 多變量解析模型作成方法、處理裝置用之多變量解析方 法、處理裝置之控制裝置、處理裝置之控制系統。 【圖式簡單說明】 第1圖係顯示關於本發明之第1實施形態的電漿處理 裝置之槪略構造的剖面圖。 第2圖係顯示第1圖所示之電漿處理裝置的多變量解 析手段之一例的方塊圖。 第3圖係顯示關於本發明之第2實施形態的處理裝置 控制系統的構造方塊圖。 第4圖係說明關於本實施形態之處理控制系統的模型 作成時的動作流程圖。 第5圖係說明關於本實施形態之處理裝置控制系統的 模型作成時之動作流程圖，爲第4圖之延續。 第6圖係說明關於本實施形態之處理裝置控制系統的 模型作成時之動作流程圖，爲第5圖之延續。 -45- (42) 100404333 $ 7圖係說明藉由關於本實施形態之 糸充的ί莫製以進行控制時的動作流程圖。 &褰元件對照表 1〇〇:電漿處理裝置，101:處理室 極’ 1〇4 :上部電極（淋浴頭）’ 105 :倡 1〇7:高頻電源，107Α:匹配器，107Β: 電氣羹測器，1 1 8 :製程氣體供給系統 器，2〇〇:多變量解析手段，201:多變量 段，2 0 5 :解析資料記憶手段，2 0 6 :運； 測·診斷·控制手段，22 1 :控制參數量 反映參數量測器，223 :裝置狀態參數量 裝置控制手段，3 0 〇 :控制系統，3 2 0 :網 處理裝置控制％ 1，102 :下部電 ^極子環型磁鐵， 瓦特計，107C : ，120 :光學檢測 ^解析程式記憶手 障手段，2 0 7 :預 測器，2 2 2 :電漿 測器，2 2 5 :處理 路，W :晶圓 • 46 -Ya ’..... Yn’ is stored in the analysis data storage means 316. Next, in step S330, the host device 310 uses the PLS method according to the program of the multivariate analysis program memory means 314 in the same manner as the first embodiment described above, with new setting data (purpose variable) Xa '' new setting The data (explanation variable) Ya 'obtains the return rank Ka' of the plasma processing apparatus 100A by the calculation means 312, and stores it in the analysis data storage means 316. Next, in step S340, these new setting data Xa ', new detection data Ya', and new return line 歹 ϋKa 'are transmitted to the plasma processing apparatus 100A through the network 320. In step S140, the plasma processing device 100A of the reference processing device receives the setting data Xa ', the detection data Ya', and the regression line Ka 'from the host device 310, and stores it as a new model. Next, as shown in Fig. 6, a model of a processing apparatus other than the reference processing apparatus, such as a plasma processing apparatus 100B, is obtained. The new model of the processing device other than the reference processing device is obtained based on the new model of the reference processing device. In processing devices other than the reference processing device, it is not necessary to newly perform plasma processing on the wafer. The specific processing will be described below. In step S530, the plasma processing apparatus 100B obtains the setting data for the regression line Kb '(the same setting data as the setting data for the return line Kb') is input by the input means 152B. This setting data is transmitted to the host device 310 through the network 320. The host device 310 receives the new setting data by the plasma processing device 100B of a processing device other than the reference processing device in step S4-10, and stores the new setting data in the analysis data storage means 3 1 6 and obtains the reception by the calculation means 3 1 2- 39- (36) (36) 200404333 The average value of each wafer for the new setting data, regard these as setting data (explanation variables) X b '..... Xn', and store them in the analysis data storage means 3 16 ° Then, in step S420, the host device 310 includes a regression line (Kb ..... Kn), a new return line (Kb '..... Kn') of a processing device other than the reference processing device, and The regression relationship (Ka) of the reference processing device and the new regression line (Ka,) of the reference processing device (for example, (Kb '/ Ka' = Kb / Ka)) are respectively obtained by the calculation means 312 (K b '..... K η'). For example, the new regression line Kb 'shown in the formula (7) of the plasma processing apparatus 100B is obtained by Kb' = Ka'Kb / Ka. Thereby, when a new regression rank of a processing device other than the reference processing device is obtained, it is not necessary to perform multivariate analysis processing such as the PLS method again, and it can be easily obtained. Next, in step S430, the host device 310 calculates a new value from the new setting data (Xb 'Xn') and the new regression line (Kb '..... Kn') based on the model shown in the above formula (7). The detection data (Yb '..... Yn') are stored in the analytical data storage means 3 1 6 and these new setting data (Xb '..... Xn'), the new regression ranks (Kb, ..... Kn ') Calculate new detection data (Yb' ..... Υη ') and send them to the corresponding plasma processing device 1 00B ..... 1 00N through the network 3 20 ° For example, in In the plasma processing device iOOB, in step S540, the host device 310 receives new setting data (Xb '..... Xu') and a new regression line (Kb '..... Kn') to calculate a new value. The test data of -40- (37) 200404333 (Yb '..... Yn') is newly memorized as shown in the above formula (8). In this way, a new model suitable for an individual processing device is obtained in a processing device other than the reference device. Next, the processing of the control system when evaluating the state based on the new model thus obtained will be described with reference to the drawings. Fig. 7 shows the operation flow of the host and the operation flow of each plasma processing device when the state of the device is evaluated based on a new model created by each electrical device. First, in a certain plasma processing apparatus 100, when an allowable variation range of a standard condition for setting data in step S610 is input, the allowable variation range is memorized. This permissible variation range is used for critical thresholds for determining whether the status is normal or abnormal. For example, it is set as a range of the maximum and minimum thresholds for the setting parameters for assigning a new model, such as the control parameters. Next, in the above-mentioned plasma processing apparatus 100, in step S620, setting data (standard conditions, for example, 例如 shown in Table 1) for processing the wafer is inputted by the input means 152, and the electrical processing is performed based on the setting data. The wafer w 'obtains the setting data and detection data measured by each wafer, and sends these setting data and detection data to the host 3 through the network 3 20. The host device 310 is processed by the above plasma in step S710. 1 〇Receive setting data and inspection data for each wafer and memorize them, and record 100 million yuan in solution. Then, the individual average 値 is obtained, and it is regarded as the material (target variable) X ', the test data (specified variable) γ, and the data storage means 3 1 6 are recorded. Next, the host device 3 1 0 in the step model is used to estimate the slurry loading device. The actual standard slurry processing material prepared by the device will be analyzed. The analysis of the capital of the device will be determined in the solution S720 -41-(38) (38) 200404333 In the process, the setting data X 'and the detection data Y' are sent to the plasma processing apparatus 100. In step S630, the plasma processing apparatus 100 receives the setting data X 'and the detection data Υ', and stores these as the actual setting data Xobs 'and the actual detection data Yobs' and stores them in the analysis data storage means 205. Next, in step S640, the actual detection data Yobs 'is applied to the new model shown in the above formula (8), and the prediction setting data Xpre' is calculated and stored in the analysis data storage means 205. Next, in step S650, the above-mentioned plasma processing apparatus 100 judges whether the setting data Xpre 'is within the allowable variation range with respect to the actual setting data Xob' to determine whether it is normal or abnormal. For example, the predicted setting data Xpre 'and the actual setting data Xobs' are judged to be normal if they are within the allowable variation range, and judged to be abnormal if they are outside the allowable variation range. When an abnormality is determined, in step S660, for example, the plasma processing apparatus 100 is stopped by the processing device control means 225, and the abnormality is notified by the display means 224 and the alarm 226. In this way, the host device 310 obtains the mean value based on the data from each plasma processing device, and performs multivariate analysis processing, so the calculation processing load of each plasma processing device can be greatly reduced. In addition, in each plasma processing device, it is not necessary to temporarily memorize a large amount of setting data or detection data obtained during plasma processing, nor does it require a multivariate analysis program, so + memory means are required for this. Thereby, the fabrication of each plasma processing apparatus can be simplified, and the manufacturing cost can be suppressed. In addition, in the second embodiment, the new model of each plasma processing device -42-(39) 200404333 is used to determine the state of the device, but this is not necessarily the case. The new model is also stored in the host device 310. Therefore, the states of the plasma processing apparatuses 100A, ..., 100N can also be determined on the main 3-10 side. In this case, when an abnormality is determined, the abnormality determination may be sent to each plasma processing device 100A ..... 10ON. Each plasma unit 100 A ..... 1 0 ON can be based on abnormality determination information, for example, by means of device control means 225 to stop processing the device, and display means alarm 226 to notify the abnormality. According to this, the device status of each plasma processing device can be monitored in the host device. Although the embodiments of the present invention have been described above with reference to the attached drawings, it goes without saying that the present invention is not limited to the embodiments. As long as it is a person in the industry, various modifications or amendments can be thought of in the scope of the patent application, and it should be understood that it belongs to the technical scope of the present invention. For example, as the setting information of the above-mentioned first and second embodiments, if a new model of the second embodiment is used to determine the state of the device when processing the wafer, the setting data by the control parameter measuring device 2 2 1 can be used. Alternatively, the data set by the input means 152 may be used. In this case, all the setting data can be measured by the parameter measuring device 2 2 1. Although the setting data measured by the control parameter 22 1 can be used, the setting data includes the amount of the control parameter. When the measuring device 2 2 1 measures, it is effective to use the input data itself. In addition, in the multivariate analysis of the above-mentioned embodiment, although it is limited to the case where the device-like information processing device is processed by the set of 224 and 310, some of these are expected to be borrowed from the input of the electrical measurement. There is an unavailable setting measured by the controlled quantity. -43- (40) (40) 200404333 Device status parameters, but the device status parameters can also be used as purpose variables or description variables. In addition, in the above embodiment, when the model is constructed, as the control parameters of the target variables, although the process gas flow rate, the gap between the electrodes, and the pressure in the processing chamber are used, as long as they are controllable parameters, they are not limited to these parameters. . In addition, as the device state parameter, although variable capacitor capacitance, high-frequency voltage, and APC opening degree are used, as long as it is a measurable parameter that displays the device state parameter, it is not limited to these. In addition, as the plasma reflecting parameters reflecting the state of the plasma, although electric data and optical data based on the plasma are used, the parameters reflecting the state of the plasma are not limited to these. In addition, although high-frequency voltages and high-frequency currents of fundamental waves and higher harmonics (up to 4 times) are used as electrical data, they are not limited to these. In addition, the output data from the means for measuring the completion of the wafer (for example, a scanning measuring instrument) assembled in the plasma processing device can also be used as the inspection data. Specifically, characteristics such as a film thickness of a film formed on a wafer, an etching amount when etching a film to be processed on a wafer, or in-plane uniformity can be used as the inspection data. In addition, in this embodiment, although the average value of the data of the plasma reflection parameters is obtained for each wafer, and using this average, the control parameters and device state parameters are predicted for each wafer, but a piece of crystal can also be used. The real-time plasma reflection parameters in the circle process, real-time prediction of control parameters and device state parameters. In the above-mentioned embodiment, although a parallel-plate-type plasma processing apparatus having a magnetic field is used, the present invention can be applied to any apparatus having control parameters, plasma reflection parameters, and / or device state parameters. -44- (41) (41) 200404333 As explained in detail above, according to the present invention, in real time, each processing device has different process characteristics or processing conditions. If a processing device is modeled, the model can be used for the same Other types of processing devices do not need to re-create models for each processing device, and can provide a multivariate analysis model creation method for a processing device that can reduce the burden of model creation and a multivariate analysis method for a processing device. Industrial Applicability The present invention can be applied to, for example, a method for creating a multivariate analysis model of a processing device such as a plasma processing device, a multivariate analysis method for a processing device, a control device for a processing device, and a control system for a processing device. [Brief Description of the Drawings] Fig. 1 is a sectional view showing a schematic structure of a plasma processing apparatus according to a first embodiment of the present invention. Fig. 2 is a block diagram showing an example of a multivariate analysis means of the plasma processing apparatus shown in Fig. 1. Fig. 3 is a block diagram showing a configuration of a processing device control system according to a second embodiment of the present invention. Fig. 4 is a flowchart showing the operation when the model of the processing control system according to this embodiment is created. Fig. 5 is a flow chart illustrating the operation flow when the model of the processing device control system of this embodiment is created, and is a continuation of Fig. 4. Fig. 6 is a flow chart describing the operation of the model of the control system of the processing device according to this embodiment, and is a continuation of Fig. 5; -45- (42) 100404333 $ 7 The figure is a flow chart illustrating the operation when the control is performed by the control system of this embodiment. & Element comparison table 100: Plasma processing device, 101: Processing chamber pole '104: Upper electrode (shower head)' 105: Advocate 107: High-frequency power supply, 107A: Matcher, 107B: Electrical tester, 1 18: Process gas supply system, 2000: Multivariable analysis means, 201: Multivariate segment, 2 05: Analysis data storage means, 2 06: Operation; Measurement, diagnosis, control Means, 22 1: Control parameter quantity reflecting parameter measuring device, 223: Device state parameter quantity device control means, 3 0: Control system, 3 2 0: Net processing device control% 1, 102: Lower electric pole ring type Magnet, Wattmeter, 107C :, 120: Optical detection ^ Analytical program to memorize hand-blocking means, 2007: Predictor, 22: Plasma detector, 2 2 5: Processing path, W: Wafer • 46-
Claims (1)
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| JP2002163869 | 2002-06-05 | ||
| JP2002168653 | 2002-06-10 |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| TW200404333A true TW200404333A (en) | 2004-03-16 |
| TWI276162B TWI276162B (en) | 2007-03-11 |
Family
ID=29738319
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| TW092115289A TWI276162B (en) | 2002-06-05 | 2003-06-05 | Multi-variable analysis model forming method of processing apparatus, multi-variable analysis method for processing apparatus, control apparatus of processing apparatus, and control system of processing apparatus |
Country Status (5)
| Country | Link |
|---|---|
| JP (1) | JP4224454B2 (en) |
| KR (1) | KR100628392B1 (en) |
| CN (1) | CN100426471C (en) |
| TW (1) | TWI276162B (en) |
| WO (1) | WO2003105210A1 (en) |
Families Citing this family (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2006001416A1 (en) * | 2004-06-29 | 2006-01-05 | Nikon Corporation | Management method, management system, and program |
| JP4643392B2 (en) * | 2005-08-24 | 2011-03-02 | 東京エレクトロン株式会社 | Operating state determination method, operating state determination apparatus, program, and storage medium for plasma processing apparatus |
| JP4164835B2 (en) * | 2005-12-20 | 2008-10-15 | 新東工業株式会社 | Projection state information estimation method, projection state information estimation device, and projection state information estimation program using a blast device |
| US7286948B1 (en) * | 2006-06-16 | 2007-10-23 | Applied Materials, Inc. | Method for determining plasma characteristics |
| US20100332010A1 (en) * | 2009-06-30 | 2010-12-30 | Brian Choi | Seasoning plasma processing systems |
| US10176279B2 (en) * | 2015-06-05 | 2019-01-08 | Uptake Technologies, Inc. | Dynamic execution of predictive models and workflows |
| US9934351B2 (en) * | 2015-11-09 | 2018-04-03 | Applied Materials, Inc. | Wafer point by point analysis and data presentation |
| JP6546867B2 (en) * | 2016-03-10 | 2019-07-17 | 東京エレクトロン株式会社 | How to adjust the processing process |
| JP7034646B2 (en) * | 2017-09-25 | 2022-03-14 | 株式会社Ｓｃｒｅｅｎホールディングス | Anomaly detection device and anomaly detection method |
| US10896833B2 (en) * | 2018-05-09 | 2021-01-19 | Applied Materials, Inc. | Methods and apparatus for detecting an endpoint of a seasoning process |
| JP6990634B2 (en) * | 2018-08-21 | 2022-02-03 | 株式会社日立ハイテク | State prediction device and semiconductor manufacturing device |
| CN112585727B (en) * | 2019-07-30 | 2023-09-29 | 株式会社日立高新技术 | Device diagnosis device, plasma processing device, and device diagnosis method |
| WO2023162856A1 (en) * | 2022-02-22 | 2023-08-31 | 株式会社Ｓｃｒｅｅｎホールディングス | Substrate processing apparatus management system, assistance apparatus, substrate processing apparatus, inter-chamber performance comparison method, and inter-chamber performance comparison program |
Family Cites Families (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH1194735A (en) * | 1997-09-22 | 1999-04-09 | Horiba Ltd | Quantitative analyzer for sample characteristic using spectrophotometry and multivariate analysis and analysis method using it |
| US6153115A (en) * | 1997-10-23 | 2000-11-28 | Massachusetts Institute Of Technology | Monitor of plasma processes with multivariate statistical analysis of plasma emission spectra |
| US6582618B1 (en) * | 1999-09-08 | 2003-06-24 | Advanced Micro Devices, Inc. | Method of determining etch endpoint using principal components analysis of optical emission spectra |
| DE60113073T2 (en) * | 2000-03-10 | 2006-08-31 | Smiths Detection Inc., Pasadena | CONTROL FOR AN INDUSTRIAL PROCESS WITH ONE OR MULTIPLE MULTIDIMENSIONAL VARIABLES |
| AU2001267913A1 (en) * | 2000-07-04 | 2002-01-14 | Tokyo Electron Limited | Operation monitoring method for treatment apparatus |
-
2003
- 2003-06-05 CN CNB038130823A patent/CN100426471C/en not_active Expired - Lifetime
- 2003-06-05 WO PCT/JP2003/007132 patent/WO2003105210A1/en active Application Filing
- 2003-06-05 JP JP2004512183A patent/JP4224454B2/en not_active Expired - Fee Related
- 2003-06-05 TW TW092115289A patent/TWI276162B/en not_active IP Right Cessation
- 2003-06-05 KR KR1020047019644A patent/KR100628392B1/en active IP Right Grant
Also Published As
| Publication number | Publication date |
|---|---|
| KR100628392B1 (en) | 2006-09-26 |
| KR20050010021A (en) | 2005-01-26 |
| JP4224454B2 (en) | 2009-02-12 |
| TWI276162B (en) | 2007-03-11 |
| JPWO2003105210A1 (en) | 2005-10-13 |
| WO2003105210A1 (en) | 2003-12-18 |
| CN100426471C (en) | 2008-10-15 |
| CN1659690A (en) | 2005-08-24 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US7505879B2 (en) | Method for generating multivariate analysis model expression for processing apparatus, method for executing multivariate analysis of processing apparatus, control device of processing apparatus and control system for processing apparatus | |
| US7289866B2 (en) | Plasma processing method and apparatus | |
| US20040235304A1 (en) | Plasma treatment apparatus | |
| US6985215B2 (en) | Plasma processing method and plasma processing apparatus | |
| US7047095B2 (en) | Process control system and process control method | |
| TW200404333A (en) | Multi-variable analysis model forming method of processing apparatus, multi-variable analysis method for processing apparatus, control apparatus of processing apparatus, and control system of processing apparatus | |
| US7490010B2 (en) | Data collection method, substrate processing apparatus, and substrate processing system | |
| EP1733325A2 (en) | Method and system for run-to-run control | |
| KR20070028204A (en) | Plasma processing apparatus and method | |
| JP2004281461A (en) | Method and apparatus of prediction of processing result | |
| TW200401179A (en) | Method of predicting processing device condition or processed result | |
| US7389203B2 (en) | Method and apparatus for deciding cause of abnormality in plasma processing apparatus | |
| TW201933016A (en) | Plasma processing apparatus and prediction apparatus of the condition of plasma processing apparatus | |
| KR102373933B1 (en) | Diagnostic system for diagnosing semiconductor processing equipment and control method thereof | |
| JP2004039952A (en) | Plasma treatment apparatus and monitoring method thereof | |
| TW202240734A (en) | Machine-learning in multi-step semiconductor fabrication processes | |
| US7630064B2 (en) | Prediction method and apparatus for substrate processing apparatus | |
| JP4220378B2 (en) | Processing result prediction method and processing apparatus | |
| TWI297904B (en) | ||
| JP2009010370A (en) | Semiconductor processing apparatus | |
| WO2024181164A1 (en) | Substrate processing system and substrate processing method | |
| US12014910B2 (en) | Method and system for adjusting location of a wafer and a top plate in a thin-film deposition process | |
| TWI677264B (en) | Rf impedance model based fault detection | |
| JP2023156153A (en) | Processed data analysis method and information processing device | |
| JPH11238723A (en) | Method and apparatus for plasma processing |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| MM4A | Annulment or lapse of patent due to non-payment of fees |