M437456 五、新型說明： 【新型所屬之技術領域】 本創作係關於一種拋光整修整器缺陷判別設備，尤指 一種節省人力、更快判別拋光墊修整器缺陷位置之設備。 【先前技術】 化學機械拋光（CMP)是平坦化常用之手段，例如用於 ^ 拋光妙晶圓。當進行拋光時，化學研磨液被添加到具特定 表面構造之拋光墊表面，藉由研磨液内的研磨粒磨削作用 使矽晶圓平坦。由於來自拋光墊、研磨液及晶圓碎屬堆積， 導致拋光墊頂部變質硬化或漸趨平坦，進而影響抛光特 性。此時便需利用修整器對拋光墊進行修整。此種修整器 基本構造包括一圓板狀基板、佈設在基板之鑽石顆粒。 因為在使用上述修整器進行拋光墊修整時有可能發 生鑽石顆粒斷落、破碎，使修整器變成為有缺陷的’因此 有必要對修整器實施一檢查步驟以確保後續使用能得到正 •埃結果。習知檢查抛光塾修整器是否發生掉鑽之作法大多 採用人工以光學顯微鏡（OM)進行視覺觀察，一旦發現有掉 鑽，便用例如油性筆之標記手段將該位置圈出再拍照，最 後再以人工比對研磨前後照片。 【新型内容】 本創作之主要目的係在提供—種拋光墊修整器缺陷 判別設備，俾能取代習知以人工針對修整器全部面積進行 M437456 檢查之作業型態，因習知作 本創作之另一目的:在：費二力又花時間° 刻m… 種拋光墊修整器缺陷 判引5又備，俾能快速判別缺陷/掉鑽之正確位置。 備包括為達的，本創作之抛光塾修整器缺陷判別設 備^括=作台、一置放座、—取像裝置、-影像處理單 二工作台包括有-工作台平面，置放座與 設於工作台平面，象處理單元電性連接於 取像裝置及顯不模組，包括一影像辨識模組、一資料儲存 器、一比對模組。資料儲存器儲存有複數參考影像資料。 上述置放座用以承載一抛光塾修整器。取像裝置與置 放^係相對可平行於工作台平面地移動用以對拋光整修 整器之不同區域產生複數試件影像。 影像辨識模組將複數試件影像處理成與複數參考影 像資料相同格式之複數試件影像資料，比對模組將複數試 件影像資料與對應之複數參考影像資料依據一比對條件分 別進行比對以判定複數試件影像資料任一者是否有缺陷情 形’比對模組更將有缺陷及無缺陷之試件影像資料有區隔 地標記’並將被有區隔地標記之試件影像資料顯示於顯示 模組。 取像裝置為任何可取像的裝置，例如照相機、或工業 用攝影機，取像裝置可包括電荷輕合元件（ChargeC〇upied Device; CCD)。為達到修整器不同區域之取像目的，可以 是取像裝f活動地設置於工作台，《者置放座彳活動地設 置於工作台，或者取像裝置及置放座都活動地設置於工作 M437456 台。 本創作之拋光墊修整器缺陷判別設備可更包括一燈 源裝置，朝拋光塾修整器照射，使獲得之影像更清晰。複 數參考影像資料與複數試件影像⑽可為灰階化格式之資 料。複數參考影像資料及複數試件影像資料更可為二值化 格式之資料’加快處理速度。複數參考影像f料與複數試 件影像:貝料亦可為彩色格式資料例如rgb彩色格式。 比對模組可更針對有缺陷/掉鑽之每一該複數試件影 像資料分料算-缺陷數4，並將該等缺陷數量顯示於顯 示模組。 【實施方式】 „旦，本創作之拋光墊修整器缺陷判別設備主要係以修整 像比對之自動化手段獲得修整器上缺陷概況如鑽石顆 粒之磨相或掉鑽等不正常佈設情形，並以視覺化結果呈現 丫 人昌 — 工作人員只要針對上述結果所指示之位置進 = 確取代了習知以人工進行全域檢測之無效率作 業方式。 丨乍甚至可用於生產品質之控制，也就是修整器成 ::時就進行檢查及缺陷分析，並非只能以研磨後修整 器為對象。 / 參考圖1〇 丄一 .2 〜本實施例之拋光墊修整器缺陷判別設備 包括一工作Αιλ m , σ 10、一置放座η、一取像裝置12、一影像處 理早 7G 3 0、一 as - » ."貝不杈組16及一燈源裝置21。工作台丨〇包括 5 M437456 有一工作台平面101，而置放座11與取像裝置12皆組設於工 作台10之工作台平面101上。置放座11用於承載一通常約略 呈圓盤狀之拋光墊修整器20。 取像裝置12用於對拋光塾修整器20之不同區域進行 取像。取像裝置12可為工業用攝影機、電荷耗合元件等。 顯示模組16具體例如為一純顯示功能之液晶顯示器，或為 同時具有顯示及輸入功能之觸控顯示器。 由於拋光墊修整器20的取像必須是針對不同區域位 置而執行’因此取像裝置12與拋光墊修整器20之間需有相 對移動，亦即取像裝置12與置放座Η需為相對可平行於工 作台平面地移動。達成此目的之作法例如置放座u採用具 二維平面移動能力之機構、取像裝置12透過可平行於工作 台平面地移動之運動機構而架設於工作台平面1(n'或上述 兩種之組合。如圖1所示，本實施例採用第三種設計，即置 放座11本身為具有二維平面移動能力之機構，同時，取像 裝置12透過一運動機構121可活動地安裝於工作台平面 1 (H ’且特別地，取像裝置12是具有6個自由度的移動能力。 上述達成平移移動目的之手段例如是使用線性滑 執、或滾珠導螺桿等常見平移機構。 燈源裝置21架設於運動機構121，其擺置係朝向拋光 墊修整器20，用於照射拋光墊修整器2〇以加強取像清晰 度。燈源裝置21可為各種常見發光器件如鎢絲燈泡、發光 二極體燈、螢光燈等。 影像處理單元30電性連接於取像裝置12及顯示模組16，主 M437456 要包括-影像辨識模組13、_資料儲存器14及一比對模組 1卜資料健存器14電性連接影像辨識模㈣與比對模組 15 ’可提供資料儲存之功能例如儲存各種影像灰階化資料 或位元RGB彩色格式資料。影像辨識模組^ 3可將取像裝 置12執订取像所取得之原色影像數位資料例如常見的靠 元RGB彩色格式資料進行處理，以符合儲存在資料儲存器 14中作為參考對像之參考影像資料。影像辨識模組η例如 將影像資料之像素轉換成1位^(此極簡之情況即為二值化 格式）4位元、或8位元等之資料格式以方便後續影像比 對步驟；或者直接利用24位元RGB彩色格式資料以進行後 續比對帛示模組i 6用於展示前述各模組配合運作之後得 出的結果資訊。 以下以一簡單範例說明本設備之使用流程。首先將一 良好之拋光墊修整器2〇(特稱為參考之拋光墊修整器）置於 置放座11，並事先規劃對拋光墊修整器區分成複數個區域 來取像分析，例如分割成4行6列，共24個區域。接著以取 • 像裝置12對參考之拋光墊修整器其中一區域取像，並將所 得之取像數位資料傳送至影像辨識模組13將其轉換成灰階 化資料（稱其為參考影像灰階化資料），然後儲存到資料儲 存器14。移動取像裝置12或置放座η以對所有其它區域執 行與上述相同之步驟’建立代表一完整的參考之拋光墊修 整器之24個參考影像灰階化資料。 一旦資料儲存器14已具有可供參考之拋光墊修整器 影像灰階化資料，便可取一待分析之拋光墊修整器2〇執行 7 M437456 2上述對參考讀光祕整器所做相同之步驟目的在得 完㈣待分析之拋光祕整器之24個試件影 階化資料。 接著比對模組15從資料儲存器14抓取相對應試件影 像灰階化請以及參相像灰階化#料進行比對所依據 之比對條件視需求可寬可嚴，藉由設定灰階化差異數值在 不同的水準而達成。 若一所針對之區域令有至少一像素位置之灰階化差 異數值超出所設定之比對條件，判定該區域為有缺陷。任 一試件影像灰階化資料若經比對判定為有缺陷情形者，則 標記為第-類’判定為無缺陷情形者，則標記為不同的第 二類。可參考圖3’最後將被標記之24筆試件影像灰階化資 料依照實際相對位置之布置、圖形化地呈現在顯示模組 16。前述有區隔的標記例如以顏色來區分，圖中顯示出代 表一待分析之拋光墊修整器之24個區域，圖中數字表示該 區域之編號以方便解讀，標記第一類之試件影像灰階化= 料呈現紅色數字，而標記第二類之試件影像灰階化資料呈 現藍色數字。如此一來，工作人員很清楚地知道只要針對 紅色數字區域進行進一步檢查確認即可，不須浪費時間在 無缺陷之區域》 所呈現的結果更設計成，當以游標C1s入特定區域， 例如編號4的區域，將更進一步顯示出細部情形，如圖4所 示。由圖4可看出，被放大顯示的區域有兩處被點標示主 知工作人員缺陷的存在位置。 而右更進步對被點標示之任一處進行點擊，還能再 Λ處/例如圖4不意出實際約有四個缺陷8G〜83存在。 。。骨：待；；析之拋光墊修整器所有區域之取像，較簡 :、方便的作法是㈣特定料進行例如圖巾的編號順 ’則所獲得之區域資料具有其可辨識性後續比對與區 域銜接顯得相當容易。但若是在較複雜之情形：以不規則 ^依特定_人序的方式進行時，為了使所得到的Μ個區域 :貝料此依實際相對位置拼凑正確’可如圖$將區域界線 Β2設定成内縮於取像界線Bl。由於圖中二取像圖案各自之 某一側像素陣财相同的灰階化數值分佈，目此可以判定 其為相銜接之區域。藉由這樣的判斷邏輯即可完成一正確 之試件影像灰階化資料銜接配置。 在較佳實施例_，比對模組可針對有缺陷/掉鑽之每一 試件影像灰階化資料分料算-缺陷數量，並將該等缺陷 數量顯示於顯示模組，如圖3所示^ :需強調的是’雖在上述實施例中所進行分析比對的資 料格式是以灰階化格式資料為對象，但當然也可 格式資料。 上述實施例僅係為了方便說明而舉例而已，本創作所 主張之權利範圍自應以申請專利範圍所述為準，而非僅限 於上述實施例0 又 【圖式簡單說明】 圖1係本創作一較佳實施例之拋光墊修整器缺陷判別設備 立體圖。 圖2係本創作一較佳實施例之拋光墊修整器缺陷判別設備 之元件之電性連接圖。 圖3係本創作一較佳實施例之顯示模組所顯示之判別結果 晝面示意圖》 圖4係圖3之拋光墊修整器局部區域放大示意圖。 圖5係具有對應銜接特徵之二取像圖案匹配示意圊。 【主要元件符號說明】 工作台平面101 取像裝置12 影像辨識模組13 比對模組15 拋光墊修整器20 影像處理單元30 游標C1 區域界線B2 工作台10 置放座11 運動機構121 資料儲存器14 顯示模組16 燈源裝置21 缺陷80〜83 取像界線B 1M437456 V. New description: [New technical field] This creation is about a polishing and finishing device defect identification device, especially a device that saves manpower and can quickly identify the defect position of the polishing pad dresser. [Prior Art] Chemical mechanical polishing (CMP) is a commonly used method for planarization, for example, for polishing a wafer. When polishing, the chemical slurry is added to the surface of the polishing pad having a specific surface texture, and the silicon wafer is flattened by abrasive grain grinding in the slurry. Due to the accumulation of polishing pads, polishing fluids and wafers, the top of the polishing pad is hardened or flattened, which affects the polishing properties. At this point, the polishing pad needs to be trimmed with a dresser. The basic structure of the dresser includes a disc-shaped substrate and diamond particles disposed on the substrate. Because the diamond particles are broken and broken when the polishing pad is trimmed using the above-mentioned dresser, the dresser becomes defective. Therefore, it is necessary to perform an inspection step on the dresser to ensure that the subsequent use can obtain positive and negative results. . It is customary to check whether the polishing 塾 dresser has been drilled. Most of the manual observations are made by optical microscopy (OM). Once the drill is found, the position is circled and photographed by, for example, an oil-based pen. Finally, Manually align the photos before and after grinding. [New content] The main purpose of this creation is to provide a polishing pad defector discriminating device, which can replace the conventional operation mode of manual M437456 inspection for the entire area of the dresser. Purpose: In: Fei Feili and time-consuming engraving m... Kind of polishing pad dresser defect judgment 5 and preparation, can quickly determine the correct position of the defect / drop drill. Included in this, the creation of the polishing 塾 dresser defect identification equipment ^ including = table, a placement seat, - image capture device, - image processing single two workbench including - workbench plane, placement and The processing unit is electrically connected to the image capturing device and the display module, and includes an image recognition module, a data storage device and a comparison module. The data storage stores a plurality of reference image data. The above-mentioned placement seat is used to carry a polishing 塾 trimmer. The image capture device is movable parallel to the table surface relative to the placement system for generating a plurality of test piece images for different regions of the polishing conditioner. The image recognition module processes the plurality of test piece images into a plurality of test piece image data in the same format as the plurality of reference image data, and the comparison module compares the plurality of test piece image data with the corresponding plurality of reference image data according to a comparison condition. In the case of determining whether any of the image data of the plurality of test pieces is defective, the sample image of the test piece having the defect and the defect is more marked than the pair module, and the image data of the test piece marked by the segment is displayed. For the display module. The image taking device is any image capturing device, such as a camera or an industrial camera, and the image capturing device may include a Charge Clamping Device (CCD). In order to achieve the purpose of image capturing in different areas of the dresser, the image capturing device may be disposed on the workbench in an active manner, and the "placement seat" is movably disposed on the workbench, or the image capturing device and the mounting seat are movably disposed on the workbench. Working M437456. The polishing pad dresser defect discriminating device of the present invention may further comprise a light source device, which is irradiated toward the polishing 塾 trimmer to make the obtained image clearer. The plurality of reference image data and the plurality of sample images (10) may be materials of a grayscale format. The multiple reference image data and the multiple test image data can be used for the data in the binarized format to speed up the processing. Multiple reference image f material and multiple sample images: The bead material can also be in color format data such as rgb color format. The matching module can further calculate the defect number 4 for each of the plurality of defective test pieces of the defective/drilled diamond, and display the number of the defects on the display module. [Embodiment] „ The visualization results are presented to the 丫人昌—the staff's position as indicated by the above results does replace the inefficient operation of the manual global inspection. 丨乍 It can even be used for the control of production quality, that is, the dresser Inspection and defect analysis are performed at the time of ::, and it is not only possible to use the post-grinding trimmer. / Refer to Fig. 1〇丄1. 2~ The polishing pad dresser defect discriminating device of the present embodiment includes a work Αιλ m , σ 10, a placement η, an image capture device 12, an image processing early 7G 3 0, an as - » . " Beibu group 16 and a light source device 21. The workbench includes 5 M437456 has a job The stage 101 is disposed, and the placement base 11 and the image taking device 12 are assembled on the table plane 101 of the table 10. The placement seat 11 is used to carry a polishing pad dresser 20 which is generally approximately disk-shaped. The image forming device 12 is configured to take an image of a different area of the polishing 塾 conditioner 20. The image capturing device 12 can be an industrial camera, a charge consuming component, etc. The display module 16 is specifically a liquid crystal display having a pure display function, or The touch display has both display and input functions. Since the image of the polishing pad conditioner 20 must be performed for different area positions, the relative movement between the image taking device 12 and the polishing pad conditioner 20 is required, that is, The image capturing device 12 and the placing frame need to be relatively parallel to the plane of the table. For this purpose, for example, the placing base u adopts a mechanism with a two-dimensional plane moving capability, and the image capturing device 12 can be parallelized to work. The movement mechanism of the table is moved on the table plane 1 (n' or a combination of the above two. As shown in FIG. 1 , the third design is adopted in the embodiment, that is, the placement seat 11 itself has a two-dimensional plane. The mechanism of the mobile capability, at the same time, the image capturing device 12 is movably mounted to the table plane 1 through a moving mechanism 121 (H' and, in particular, the image capturing device 12 is capable of moving with 6 degrees of freedom. The above-mentioned means for achieving the purpose of translational movement is, for example, a linear sliding mechanism, or a common translation mechanism such as a ball lead screw. The light source device 21 is mounted on the moving mechanism 121, and its placement is directed toward the polishing pad conditioner 20 for illuminating the polishing pad. The image processing unit 30 is electrically connected to the image capturing device 12 and the light source device 21 can be used for various common light emitting devices such as a tungsten light bulb, a light emitting diode lamp, a fluorescent lamp, and the like. The display module 16 and the main M437456 include an image recognition module 13, a data storage device 14, and a comparison module 1 and a data storage device 14 electrically connected to the image recognition module (4) and the comparison module 15'. The data storage function is, for example, storing various image grayscale data or bit RGB color format data. The image recognition module ^3 can process the primary color image digital data obtained by the image capturing device 12, for example, the common RGB color format data to conform to the reference stored in the data storage 14 as a reference object. video material. The image recognition module η converts, for example, the pixels of the image data into a 1-bit ^ (in the case of a minimalization, a binary format) 4-bit, or 8-bit data format to facilitate subsequent image comparison steps; or The 24-bit RGB color format data is directly used for the subsequent comparison display module i 6 for displaying the result information obtained after the cooperation of the foregoing modules. The following is a simple example to illustrate the flow of the device. First, a good polishing pad conditioner 2 (referred to as a reference polishing pad conditioner) is placed in the placement seat 11, and the polishing pad dresser is planned in advance to divide into a plurality of areas for image analysis, for example, into 4 rows and 6 columns, a total of 24 areas. Then, the image capturing device 12 takes an image of one of the reference polishing pad conditioners, and transmits the obtained image data to the image recognition module 13 to convert it into grayscale data (referred to as reference image gray). The data is stored and then stored in the data store 14. The image capture device 12 or the placement base η is moved to perform the same steps as described above for all other regions to create 24 reference image grayscaled data representing a complete reference polishing pad conditioner. Once the data storage 14 has the polishing pad imager grayscale data available for reference, a polishing pad conditioner 2 to be analyzed can be performed. 7 M437456 2 The same steps as described above for the reference light reader are performed. Objective To obtain (4) the shadowing data of 24 specimens of the polishing secretizer to be analyzed. Then, the comparison module 15 grabs the grayscale of the corresponding test piece image from the data storage device 14 and compares the reference grayscale material. The comparison condition can be widened according to the demand, and the gray scale is set by the grayscale The difference value is achieved at different levels. If a region is targeted such that the grayscale difference value of at least one pixel position exceeds the set comparison condition, the region is determined to be defective. If the grayscale data of any of the test pieces is judged to be defective by comparison, the one marked as the first class is judged to be defect-free, and the second class is marked as different. Referring to Fig. 3', the 24 sample image grayscale materials to be marked are finally graphically presented on the display module 16 according to the arrangement of the actual relative positions. The aforementioned distinguishing marks are distinguished, for example, by color. The figure shows 24 areas representing a polishing pad conditioner to be analyzed. The numbers in the figure indicate the number of the area for easy interpretation, and the image of the first type of test piece is marked. Grayscale = material is presented with a red number, while grayscale data for the specimen image of the second category is presented with a blue number. In this way, the staff is very clear that as long as the red digital area is further checked and confirmed, the result of not wasting time in the defect-free area is more designed, when the cursor C1s is entered into a specific area, such as numbering. The area of 4 will further show the details, as shown in Figure 4. As can be seen from Fig. 4, the enlarged area shows two locations where the points indicate the presence of the defect of the worker. The right progress is made to click anywhere in the point mark, and it can be further circulated/for example, Figure 4 does not mean that there are actually about four defects 8G~83. . . Bone: Wait; Analyze the image of all areas of the polishing pad dresser, simpler: Conveniently, (4) the specific material is carried out, for example, the number of the towel is compliant, and the obtained regional data has its identifiability and subsequent comparison. It is quite easy to connect with the area. However, if it is in a more complicated situation: in the case of irregular ^ according to the specific _ person order, in order to make the obtained area: the material is assembled correctly according to the actual relative position 'can be set as shown in Fig. $ regional boundary Β 2 Inwardly shrinks to the image boundary B1. Since the gray-scaled value distribution of the same pixel of each of the two image-taking patterns in the figure is the same, it can be determined that it is the connected region. With such a judgment logic, a correct configuration of the grayscale data of the test piece image can be completed. In the preferred embodiment, the comparison module can calculate the number of defects for each grayscale data of the defective/drilled image, and display the number of defects in the display module, as shown in FIG. Shown ^: It should be emphasized that 'the data format of the analysis comparison performed in the above embodiment is the gray-scale format data, but of course the format data. The above-mentioned embodiments are merely examples for convenience of description, and the scope of the claims claimed herein is based on the scope of the patent application, and is not limited to the above-mentioned embodiment 0. A perspective view of a polishing pad conditioner defect discriminating device of a preferred embodiment. Fig. 2 is an electrical connection diagram of the components of the polishing pad conditioner defect discriminating device of the preferred embodiment of the present invention. 3 is a schematic diagram showing the result of the display of the display module of the preferred embodiment of the present invention. FIG. 4 is an enlarged schematic view showing a partial area of the polishing pad conditioner of FIG. Figure 5 is a schematic representation of the matching of two image patterns with corresponding engagement features. [Main component symbol description] Workbench plane 101 Image capture device 12 Image recognition module 13 Comparison module 15 Polishing pad conditioner 20 Image processing unit 30 Cursor C1 Area boundary B2 Work table 10 Placement seat 11 Motion mechanism 121 Data storage 14 display module 16 light source device 21 defect 80~83 image boundary B 1