1245944 玖、發明說明： C 日月戶斤4支々貝】 發明領域 本發明係有關於一種面板貼合方法，更詳而言之，係 5有關於一種一面進行高精度之對位—面進行貼合之面板貼 合方法及裝置。 C先前3 習知技術 一般用以作為顯示裝置之液晶面板裝置，係液晶驅動 1〇用之薄膜電晶體基板（TFT面板）與濾色基板（CF面板）貼 合，且其間密封有液晶之結構。前述液晶面板裝置之製造 步驟，係使用面板貼合裝置將前述TFT面板與(：^面板透過 密封黏著材貼合。 如前述之液晶顯示裝置用之TFT面板與cF面板之貼 15合，係需要數V m左右之高對位精度，因而於一面加壓2片 面板一面進行貼合之步驟中，必需高精度地進行面板組之 對位’故’要使用具有對位機能之面板貼合装置來進行面 板之貼合處理。 藉習知之面板貼合裝置進行面板之貼合處理步驟時， 2〇係將2片如TFT面板與CF面板之面板，分別安裝於對向之上 下之平板’並支持前述2片面板使其成為對向之狀態。接 著’ §使上側之平板相對於下側之平板朝垂直方向移動而 對面板組加壓來進行貼合時，使前述下側之平板相對於上 側之平板平行移動來進行該面板組之對位。 1245944 第1圖係顯示習知面板貼合步驟之一例之流程圖。又， 第2圖係模式地顯示第1圖之面板貼合步驟中兩片面板相對 地移動之圖。 百先，將兩片面板安裝於上下之平板使其成為對向之 狀態（步驟Sl)，並藉辨識攝影機讀取對準標誌（步驟S2)。其 中，一般習知之貼合裝置中，上平板可朝垂直方向（2軸方 向）移動，而下平板可朝水平方向（X、Y方向及旋轉）方 向方疋轉）移動。又，對準標誌係設置於各面板之一部份 部）。 10 15 接著，判定上下面板之對準標誌是不是於預定之範圍 内（步驟S3)。即，判定上下面板是不是已以預定精度對位， 尚未以預定精度對位時，使下平板朝χ、γ、及Θ方向移動 〇=S4) ’ i進行再度對位。以上步驟係進行粗略對位(粗 略對準)之步驟。前述粗略對位之步财進狀對位，係使 上下面板之錯位為例如5# m以下。 結束前述粗略對位後，再進行細微對位(細微對準）。细 微對位係與前述步驟S2〜以相同之處理（步驟S5〜s?)，且所 進行之對位係使上下面板之錯位為例如以下。 呵點鏈綠係表示步驟S1中下面板之位置。 其中使上面板於z方向上之位置為P1時，於ρι之位置之狀態 下移動下面板來進行粗略_，上下面板成為排列整齊之 狀態。即，於Μ狀態下，藉施行步驟如4之處理進行粗 略對位，且藉施行步驟S5〜S7之處理進行細微對位。藉施行 粗略對位與細《位，τ面板如P1巾之實叙箭號標示般 20 1245944 對位之狀能） 面板已呈排列整齊之狀態（已進行細微 動，=位結束時，使前述上平板相對於下平板垂直移 加壓為=?板成為重疊之， 後，：取㈣剛開始先加壓至例如3Gkg(步驟S8)。加壓 驟si上下面板之對準標認(步驟S9)，並確認對位精度(步 且再产進亡ΙΙΓ之對位偏離時’則移動下平板(下面板) 再度進仃上面板與下面板之細微對位。 10 15 、〈後處理回到步驟S8，並實施步驟S9及S10之處理， 於trsiG判定已以財精度對位時，則判定之前之加壓是 曰最；加[(步驟S12)。若非最終加Μ，則回到步驟§8並 再度進行步驟S8〜S12。 第1圖所不之例中，令最終加壓為100kg，並於加壓中 ㈣加壓之位置進行細微對位。即 ，於最初之步 之處理先加壓至3Qkg，並於下—步驟％之處理加壓至 6〇kg，且於再下一步驟S8之處理加壓至i〇〇kg。 於第2圖中，於加壓至30kg之狀態下，上面板於z軸方 向上之位置係P2之位置。上面板由ρι移動至？2時，如虛線 之前唬所示般於X、γ方向發生錯位，前述錯位主要是由用 以移動上平板之移動機構造成的。 因此’於P2之位置中移動下面板來進行上下面板之細 微對位（最初之步驟S11之處理）。接著，對上下面板施加60kg 之外加壓力。此時，上面板之位置為P3。上面板*p2移動 至P3之位置時亦同樣發生錯位，因此，於p3之位置中亦要 20 進行細微對位（第2次之步驟S11之處理）。 然後’對上下面板施加最終壓力之1〇〇kg之外加壓力。 此時，上面板之位置於PF。上面板由P3移動至？17之位置時 亦同樣發生錯位，因此，於PF之位置中亦要進行細微對位 (第3次之步驟su之處理）。 反覆操作如前述之加壓與對位後，於步驟S12判定之前 之加壓為最後加壓時，測定上下面板之對準標誌之錯位量 / & S13)’並結束面板之貼黏處理。依據前述之貼合步驟， 可將兩片面板以〇.5vm〜i.5vm之對位精度貼合。 如韵述’第1圖所示之習知之面板貼黏步驟中，加壓步 驟與對位步驟係交互地分別進行，因此貼合步驟所需之處 理時間增長。 V因此，有先預測貼合步驟中之錯位量，並事先使下面 板偏離預測之錯位量之方法（參照例如專利文獻1)。 第3圖係使用使面板偏離預測之錯位量之方法進行面 板貼合步驟之一例之流程圖。又，第4圖係模式地顯示第3 圖之面板貼合步驟中兩片面板相對移動之圖。 第3圖所示之貼合步驟中，將兩片面板以對向之狀態安 裝於上下之平板(步驟S21)，並藉辨示攝影機讀取對準標誌 (步驟S22)。步驟S21中，係以上下面板盡可能地靠近之狀 怨配置’上下面板間之距離例如為2//m左右。 接著’判定上下面板之對準標誌是不是於預定之範圍 内（步驟S23)。即，判定上下面板是不是已以預定精度對位， 尚未以預定精度對位時，使下平板朝X、γ、及Θ方向移動 1245944 (步驟S24)，並進行再度對位。 步驟S24之處理，並不只是使下面板與上面板對位，是 要使下面板移動至已將加壓時之錯位量納入考慮之位置 上。且，於依據預測位置進行對位之對位結束後，對上下 5 面板施加例如300kg之外加壓力（步驟S25)。此時之加壓係由 0〜300kg連續進行。 即’步驟S24之處理，如第4圖所示，係於pi之位置中 使下面板如實線之箭號所示移動至以虛線表示之面板之位 置上。该以虛線表示之面板之位置’係已事先將於加壓時 1〇 會產生之錯位量納入考慮之位置。因此，對上下面板連續 地施加外加壓力，並達到300kg之最終外加壓力時，即，上 面板如第4圖之虛線之箭號所示到達最終位置pF時，上下面 板之對位結束。 除了以上之例以外，尚有於加壓兩片基板時，同時進 5彳于垂直方向之加壓動作與水平方向之對位動作之方法。 即’一面加壓一面檢測面板組之錯位，並基於檢測結果於 加壓中修正錯位之方法。（例如，參照曰本專利文獻2及3。） 【曰本專利文獻1】 曰本專利公開公報HM0508號 20 【曰本專利文獻2】 曰本專利公開公報2000-25 8746號 【曰本專利文獻3】 日本專利公開公報2002-40398號 兔明欲解決之課題 1245944 前述第i圖所示之面板貼合步财，係一面使面板組相 對移動一面階段性地進行加壓來進行貼合。因此，即使於 第^階段（第2圖之P1之位置）中兩片面板之中心正確地對位 (對準），但只要於其次之階段（第2圖之打、朽、及柙之位置） 日守產生中心位置偏離，每階段仍要進行對位處理。 因此，於加壓步驟與對位步驟交互地進行之情況下， 面板貼合步驟中進行對位之時間增長，面板貼合步驟所需 之時間因而增加。 又，每階段之對位中，係於加壓兩片面板之狀態（例如 30kg、60kg、l〇0kg)下使面板組相對移動。此時，會對介 :兩片面板間之密封黏著材產生剪切應力，造成該密封黏 著材有變形或損傷之虞。密封黏著材損傷時，應該會密封 於兩片面板間之液晶漏出’使液晶顯示裝置無法使用。 又，前述第3圖所示之面板貼合步驟中，兩片面板亦於 10 15 進行加壓步驟時相對移動，因此黏著材同樣地㈣m 損傷之虞。然而，此時之面板的相對移動並非對位動作， 而是加壓機構之誤差所引起的。 且於加壓中測出錯位並修正之方法，係於測出某程 度之錯位時進行錯位修正。欲於短加壓時間内進行錯位修 正時’必需使錯位量之容許範圍某程度地擴大，並使自加 壓開始^加壓結束前進行之錯位修正之次數減少。然而， 使錯位里之各許範圍擴大時，加壓中之面板之相對移動量 (錯位修正量)增加，密封黏著材會有損傷之虞。又，欲避免 密封黏著材損傷時，必需增加錯位修权讀，加壓步驟 20 1245944 之時間因而增長。 【發明内容】 發明概要 个设巧係亏重丽述觀點製 可成者，且目的在於提供〜插 了縮短於面板貼合步驟中 種 ,, 订之對位時間，且可減少密封 黏者材之損傷之面板貼合方法及裝置。 山封 用以解決前述課題之手段 10 15 為了解決前述課題，本發明係提供一種面板貼合方 ^係使兩片面板隔著黏著材加壓㈣合而成為疊合之狀 L，又，係於前述兩片面板之面板面對向之狀態下 兩片面板，以隔著前述黏著材朝與 面板面垂直之方向以預定之外加壓力加壓前述兩片面 板組’且於達到前述預定之外加壓力之前，至少使前逃兩 片面板之其中-者朝與前述面板面平行之方向移動相當於 事先預測並设定之預測錯位量之移動距離，並且於到達前 述預定之外加壓力時，於前述兩片面板已對位之狀態下結 束加壓。 依據如述本發明之面板貼合方法，/係於進行面板之貼 Β日守，一面加壓前述上面板一面使下面板移動，g於加壓 之同％除去2片面板間之錯位，因而可使兩片面板之相對移 動里降低至最小限度j因此，可防止設置於面板間之密封 黏著材等變形或損傷，而可實現具有高精度之對位機能， 且可進行高可靠性之貼合之面板貼合裝置。 本發明之面板貼合方法，亦可於開始加壓至結東之前 1245944 暫時停止前述加壓，並測出前述面板組於停止前之許位 量，且於該錯位量超過預定之錯位量時，至少使前述兩片 面板中之其中一者朝與前述面板面平行之方向移動以修正 該錯位量，且於前述錯位量修正結束後再度開始加壓。 5 或者，可於開始加壓至結束之前，測出前述面板組之 錯位量，並於該錯位量超過預定之錯位量時，一面連續地 加壓’一面使前述兩片面板中之至少一者朝與前述面板面 平行之方向移動以修正該錯位量。 又’前述之面板貼合方法，係以使已測出之前述錯位 10里與前述兩片面板之距離具有相關性並作為資訊加以儲 存’用以設定之後之面板貼合步驟中之預測錯位量為佳。 或者’亦可使已測出之錯位量與加壓量具有相關性並作為 貧訊加以儲存，用以設定之後之面板貼合步驟中之預測錯 位量。 且，本發明之面板貼合方法，可於測出之前述錯位量 較，據前㈣存之錯位資訊求出之錯位量超出狀之值以 上時，發出用以通知發生異常狀況之警報。 20 、’本發明係提供—種面板貼合裝置，該面板貼合裝 =以使兩片面板隔著黏著材加I與貼合而成為疊合之 機。包合有上T平板、上平板移動機構、桌檯、攝影 板者，^Τ平㈣相分料«述兩片面 向移動動機構係用以使前述上平板朝垂直方 夕動者’月II述桌檯係用以使前述 者，該攝影機係用以對分別設置2朝水平方向移動 、月J迷兩片面板之對準標 12 1245944 誌進行影像辨識者，而前述控制裝置，係用以控制前述上 平板移動機構之動作，並控制前述桌檯之移動以藉由前述 攝景^機得到之景〉像辨識結果進行前述面板組之對位，且前 述控制裝置係一面使前述上平板移動機構動作且加壓前述 5面板，一面控制鈾述桌檯之移動以使前述下平板移動相當 於事先預測並設定之預測錯位量之移動距離。 依據前述本發明之面板貼合裝置，於進行面板之貼合 曰守，一面加壓前述上面板一面使下面板移動，而於加壓之 同時除去2片面板間之錯位，因而可使兩片面板之相對移動 1〇量降低至最小限度。因此，可防止設置於面板間之密封黏 著材等變形或損傷，而可提昇面板貼合裝置之可靠性。又， 可儘可能地縮小對位動作中之面板之移動距離，而可縮短 貼合處理所必需之時間。因此，可提升面板貼合裝置之性 能0 15 20 立本發明之面板貼合裝置中，前述控制裝置係以包括記 憶裝置及預測機構為佳，該記憶機構係用以將已對應前述 垂直方向位置測出之前述下平板之水平方向位置 〜、貝λ加以儲存者，而賴機構係用以將前述儲 前述錯位量之預測者。χ，前述控制裝‘ =構：警報機構係於已測出之前述錯位量較根據 可‘:::τ求出之錯位量超出預定之值以上時， 出用以通知發生異常狀況之警報者。 t實施方式3 較佳實施例之詳細說明 13 1245944 接著，針對本實施形態，與圖示一併作說明。 首先’針對使用本發明之面板貼合方法之面板貼合裝 置，一面參照第5圖一面說明。第5圖係使用本發明之面板 貼合方法之面板貼合裝置之截面圖。 5 第5圖所示之面板貼合裝置1〇,係用以將構成液晶顯示 裝置之TFT基板(TFT面板)與渡色基板(CF面板)力ϋ壓並黏著 之裝置。TFT面板形成有液晶之驅動電路，且藉已封入打丁 面板與CF面板之間之液晶，將資訊顯示於(^面板上。 面板貼合裝置10具有上平板14及下平板18，上平板14 10係安裝CF面板丨2且可朝垂直方向（Z軸方向）移動者，而下平 板18係安裝TFT面板16且可朝水平方向移動者。上下平板 14、18係配置於由上部室20與下部室22構成之室内。於上 部室20與下部室22密閉時室可成為真空狀態，且室内可設 定成預定之真空壓力。 15 上部室20連接支持構件24，且支持構件24係藉氣筒等 構成之往返移動機構而可朝垂直方向（2軸方向）移動。室之 開關係藉由上部室20上下移動進行。 上平板14連接貫通上部室2〇之壁且延伸至室内之支持 構件26。藉由上平板移動機構28使支持構件26朝垂直方向 20 (Z軸方向）移動，可使室内之上平板14朝垂直方向（Z軸方向） 移動。 此外’上平板移動機構28可由如圖示之滾珠螺桿機構 或線型馬達等週知之直線運動機構構成。第5圖所示之例 中’以驅動滾珠螺桿之伺服馬達3〇作為驅動源而驅動上平 1245944 板移動機構28，使上平板14朝垂直方向（z軸方向）移動。 配置於下部室22内之下平板18，連接貫通下部室22之 底部且延伸之支持構件32。支持構件32固定於χγ0桌檯 34，下平板18可於室内朝水平方向移動。即，藉使χγθ桌 5檯34於水平面内朝X方向、υ方向、及θ方向（旋轉方向）移 動，可使下平板於室内朝X方向、γ方向、及θ方向（旋轉方 向）移動。由於含有驅動源之ΧΥΘ桌檯34可由週知之移動機 構構成，所以省略其詳細說明。 前述之面板貼合裝置中，上平板移動機構28安裝於牢 10固之框架36，且用以導引上平板移動機構28於垂直方向上 移動之導引機構38安裝於框架36。導引機構38係導引上平 板移動機構28使上平板14可精度良好地垂直移動之機構， 然而因機械之尺寸誤差或安裝誤差等影響，並不會是完全 垂直之方向。 15 即，上平板14相對於下平板18之垂直移動，並非完全 垂直，當上平板14朝垂直方向移動時，會相對於下平板18 朝水平方向（X、Υ、0方向）稍許偏離。為了修正如此之偏 離，下平板18固定於ΧΥΘ桌檯34，而可朝水平方向移動。 即’藉驅動ΧΥΘ桌檯34，進行上下面板安裝於上下平板14、 20 18時之對位，及於加壓時產生之錯位之修正。 於上平板移動機構28設置有於對位時作為感測器之影 像辨識攝影機40，影像辨識攝影機40透過設置於上部室2〇 之透明窗與設置於上平板之14之貫通開口，對分別設置於 CF面板12及TFT面板16之對準標諸進行影像辨識。即，— 15 1245944 面藉影像辨識攝影機40對兩者之面板12、16之對準標誌進 行影像辨識，一面進行CF面板及TFT面板之對位。此外， 為了要辨識對準標誌於X、Y、0方向之偏離，影像辨識攝 影機40係以設置在2處以上不同之位置為佳。 5 又’上平板移動機構28設置有作為載重感測器之負荷 感知裔42。負荷感知器42可測知安裝於上平板14之CF面板 12相對於安裝於下平板18之TFT面板16按壓時之載重（即， 外加壓力）。 其中’用以使上平板14垂直移動之驅動源之伺服馬達 10 30之動作係由控制裝置44控制。又，用以使下平板以朝水 平方向私動之χγθ桌檯34亦由控制裝置控制。控制裝置 44接收由珂述影像辨識攝影機40之影像辨識結果顯示之對 位貝讯，及由負荷感知器42之外加壓力之檢測結果顯示之 外加壓力貧訊。因此，控制裝置44根據來自前述影像辨識 15攝影機4〇之對位資訊及來自負荷感知 器42之外加壓力資 訊，控制上平板移動機構28及ΧΥΘ桌檯34之動作。 又，控制裝置44為了如後述般學習對位及錯位修正， 宜具有用以儲存各種資訊之儲存器（記憶機構）、及根據該等 貝Α預測錯位$演鼻錯位修正量等之處理器（預測機構）。 20 且，控制裝置44宜具有警示器（警報機構）， 該警示器係於 已測出之錯位量與實際之錯位量之差超出預定之值以上時，可 判定裝置發生異常狀況而發出警報者。 此外，岫述第5圖所示之例中，於上平板14安裝有 面板12’亚於下平板18安裝有TF丁面板16，然而亦可於上平 16 1245944 板14安裝TFT面板16，而於下平板18安裝CF面板12。以下， 安裝於上平板14之面板稱作上面板，安裝於下平板μ之面 板則稱作下面板。 且，於貼合TFT面板及C!F面板時，於下面板之周邊 5部框狀地設置密封黏著材，並供給液晶至其内側。利用使 上下面板藉由密封黏著材加壓並黏著，於上下面板間密封 液晶。可藉使用例如UV硬化型黏著劑，於貼合處理後照射 紫外線使其完全硬化，作為密封黏著材。 接著，針對本發明之第1實施例之對位方法，一面參照 10 弟6圖至弟8圖一面作說明。 第6圖係顯示使用本發明之第丨實施例之面板貼合方法 之面板貼合步驟之流程圖。又，第7圖及第8圖係模式地顯 不第6圖所示之面板貼合步驟中兩面板相對地移動之圖。 第6圖所示之面板貼合步驟，係使用第5圖所示之面板 15貼合裝置10，首先使上下面板於對向之狀態下安裝於上下 平板14、18(步驟S31)，且藉影像辨識攝影機4〇讀取設置於 上下面板12、16之對準標誌（步驟S32)。 接著，判定上下面板之對準標誌是不是已於預定之範 圍内（步驟S33)。即，判定上下面板是不是已以預定之精度 2〇對位。尚未以預定之精度對位時，使下平板18被、γ、= 方向移動(步驟S34)，並進行再度對位。以上之步驟係粗略 對位(粗略對準)之步驟。粗略對位步_進行對位，使上下 面板之錯位為例如5//m以下。 粗略對位結束後，再進行細微對位(細微對準）。細微對 17 !245944 位係與前述之步驟S32〜S34相同之處理，且進行對位使上下 面板之錯位為例如0.5//m以下。 細微對位結束時，使上平板14相對於下平板16垂直移 動，並於兩片面板重疊之狀態下加壓。此時，令最終加壓 5為1001^時，首先先加壓至例如30kg(步驟S38)。如後述，該 步驟係於垂直移動上平板14使上面板對下面板加壓時，使 下平板18(下面板)移動以追隨上面板之錯位。即，預測上面 板之水平方向之移動，且亦使下面板於同時朝同一方向移 動（預測移動）。 於加壓後讀取上下面板之對準標誌（步驟39)，並判定上 下面板之錯位是不是於容許範圍内（步驟S4〇)。當上下面板 之錯位超出容許範圍時，驅動ΧΥΘ桌檯34使下平板18(下面 板)移動，並再度進行上面板與下面板之細微對位。 然後，處理回到步驟S38，並進行步驟S39與步驟S4〇 Η之處理。於步驟S40中判定上下面板之錯位已於容許範圍内 時，則判定之前進行之加壓是不是最終加壓（步驟42)。不是 最終加壓時，回到步驟S38，並再度進行步驟S38〜步驟以2。 第6圖所示之例中，令最終加壓為職§，並於加壓中 在30kg與60kg加壓之位置進行細微對位。即，於最初之步 20驟S38之處理先加壓至撤g，並於下_步驟s38之處理加壓 至60kg，且於再下一步驟S38之處理加壓至臟§。於加壓 至100kg時，判定該加壓為最終加壓。 第7圖及第8圖係模式地顯示前述面板貼合步驟中上下 面板之移動之圖。第7圖係顯示下面板之_移動如預期地 1245944 對應上面板之錯位之例。另外，第8圖則顯示使下面板相對 於上面板之錯位而預測移動，然而由於產生誤差，所以於 加壓途中僅能進行些許修正之例。 於第7(a)圖中，於加壓至30kg之狀態下，上面板於z軸 5 方向上之位置係P2之位置。上面板於垂直方向上由pi移動 至P2時，如虛線之箭號所示於X、Y方向上產生錯位。該錯 位主要是由用以移動上平板14之上平板移動機構28之機械 誤差造成的。為了不要產生前述錯位，當使上面板由P1移 動至P2時’如第7(b)圖所示，同時使下面板移動上面板於水 10 平方向上之錯位量（最初之步驟S38之處理）。 第7圖所示之例中，於P2之位置中，判定面板之錯位已 於容許範圍内（步驟S40之YES)。因此處理回到S38,對上下 面板施加60kg之外加壓力。外加壓力達到6〇]^時，上面板 之位置在P3。上面板由P2移動至P3之位置時，亦同樣發生 15錯位。因此，上面板由Μ移動至P3之位置時（即，外加壓力 由30kg增加至60kg時，亦使下面板預測移動（第2次之步驟 S38之處理）。 接著，對上下面板施加最終壓力i100kg之外加壓力。 外加壓力達到l〇〇kg時，上面板之位置在PF。上面板由P3 20移動至PF之位置時’亦同樣發生錯位。因此，外加壓力由 6〇kg增加至丨⑽以時，亦使下面板預測移動（第3次之步驟 S38之處理）。 反覆操作如以上之加壓後，於步驟S42判定之前之加壓 為最終加壓時，測定上下面板之對準標誌之錯位量（步驟 19 1245944 S43)，面板之_處理結束。藉如 面板™〜一兩片 依據如前述之t, 之同時使下面板朝水平方=於上面板朝水平方向移動 進行時亦不會產4 因此面板即使於加壓正 穷封黏著材僅-錯位。故’夾置於上下面板之間受壓之 :材i生jc/^直方向之外加壓力，可防止於密封黏 者材產水平方向之剪切應力而損傷其。 10 151245944 发明 Description of the invention: C sun moon household catty 4 scallops] FIELD OF THE INVENTION The present invention relates to a panel bonding method. More specifically, the system 5 relates to a high-precision alignment on one side-surface Laminated panel laminating method and device. C Previous 3 The conventional technology is generally used as a liquid crystal panel device for a display device, which is a structure in which a thin film transistor substrate (TFT panel) and a color filter substrate (CF panel) for liquid crystal driving 10 are bonded, and a liquid crystal is sealed therebetween. . The manufacturing steps of the aforementioned liquid crystal panel device are to use a panel bonding device to bond the aforementioned TFT panel and (: ^ panel through a sealing adhesive material. As described above, the TFT panel used for the aforementioned liquid crystal display device and the cF panel are bonded together, which is required. High registration accuracy of several V m. Therefore, in the step of laminating two panels while pressing one side, it is necessary to perform the alignment of the panel group with high accuracy. Therefore, a panel bonding device with an alignment function must be used. To perform the panel bonding process. Using the conventional panel bonding device to perform the panel bonding process, 20 is to install two panels, such as a TFT panel and a CF panel, on the upper and lower flat panels. Support the aforementioned two panels so that they face each other. Then '§ Move the upper flat plate vertically to the lower flat plate to press the panel group for bonding, and make the lower flat plate relative to The upper plate is moved in parallel to perform alignment of the panel group. 1245944 Figure 1 shows a flowchart showing an example of the conventional panel bonding steps. Also, Figure 2 shows the panel of Figure 1 in a pattern. Figure of the two panels moving relative to each other in the combined steps. Baixian, the two panels are mounted on the upper and lower flat plates to make them face each other (step S1), and the alignment mark is read by the identification camera (step S2). Among them, in the conventional bonding device, the upper plate can move in the vertical direction (two-axis direction), and the lower plate can move in the horizontal direction (X, Y direction and rotation). The alignment mark is provided on a part of each panel). 10 15 Next, it is determined whether the alignment marks of the upper and lower panels are within a predetermined range (step S3). That is, it is determined whether the upper and lower panels have been aligned with a predetermined accuracy, and when the upper and lower panels have not been aligned with a predetermined accuracy, the lower plate is moved in the directions of χ, γ, and Θ 〇 = S4) ′ i to perform alignment again. The above steps are steps for rough alignment (rough alignment). The foregoing rough alignment step-by-step alignment makes the displacement of the upper and lower panels to be, for example, 5 # m or less. After the rough alignment is completed, fine alignment (fine alignment) is performed. The fine alignment is performed in the same manner as in the above steps S2 to S5 (steps S5 to s?), And the alignment is performed so that the upper and lower panels are displaced, for example, as follows. The green dot chain indicates the position of the lower panel in step S1. When the position of the upper panel in the z direction is P1, the lower panel is moved in the state of ρ to make a rough _, and the upper and lower panels are arranged neatly. That is, in the M state, rough alignment is performed by performing the processing such as step 4, and fine alignment is performed by performing the processing of steps S5 to S7. By performing rough alignment and fine alignment, the τ panel can be aligned as the actual arrow of P1 towel 20 1245944. The panel has been in a neatly arranged state (fine movement has been performed, = at the end of the alignment, make the aforementioned The vertical pressure of the upper plate relative to the lower plate is equal to that of the? Plate. After that, the plate is first pressed to 3 Gkg, for example (step S8). The alignment step of the upper and lower panels is marked (step S9). ), And confirm the alignment accuracy (when the misalignment of the reproduced ΙΓΓ is misaligned, then move the lower plate (lower panel) to enter the fine alignment of the upper and lower panels again. 10 15 、 <Post processing returns Step S8, and implement the processing of steps S9 and S10. When trsiG determines that it has been aligned with financial accuracy, it determines that the previous pressure is the highest; add [(step S12). If it is not the final addition of M, return to step § 8 and then perform steps S8 to S12 again. In the example shown in Fig. 1, the final pressurization is 100 kg, and the position is slightly aligned during the pressurization. That is, the treatment in the first step is added first. Press to 3Qkg, and press the treatment in the next step% to 60kg, and in the next step S8 Press up to 100 kg. In the second figure, in the state of pressurizing to 30 kg, the position of the upper panel in the z-axis direction is the position of P2. When the upper panel is moved from ρ to? 2, as before the dotted line A misalignment occurs in the X and γ directions as shown, the aforementioned misalignment is mainly caused by the moving mechanism used to move the upper plate. Therefore, 'moving the lower panel in the position of P2 to perform the fine alignment of the upper and lower panels (the initial step) Treatment of S11). Next, apply 60kg of pressure to the upper and lower panels. At this time, the position of the upper panel is P3. When the upper panel * p2 is moved to the position of P3, the same displacement occurs, so it is necessary to position p3. 20 Perform fine alignment (the processing of step S11 for the second time). Then, apply pressure to the upper and lower panels in addition to 100 kg of the final pressure. At this time, the position of the upper panel is at PF. Is the upper panel moved from P3 to? The position 17 is also misaligned. Therefore, it is necessary to perform a slight alignment in the position of PF (the processing of the third step su). After repeated operations such as the aforementioned pressure and alignment, before the determination in step S12 When pressurized is the last pressurization ， Measure the amount of misalignment of the alignment marks on the upper and lower panels / & S13) 'and finish the sticking process of the panels. According to the aforementioned bonding steps, the two panels can be pasted with the alignment accuracy of 0.5vm ~ i.5vm. In the conventional panel sticking step shown in the rhyme '1, the pressing step and the alignment step are performed separately and separately, so the processing time required for the bonding step increases. Therefore, there is a prediction A method of shifting the amount of dislocation in the bonding step and deviating the lower panel from the predicted amount of dislocation in advance (see, for example, Patent Document 1). Fig. 3 is an example of a panel bonding step using a method of displacing the panel from the predicted amount of dislocation. flow chart. Fig. 4 is a diagram schematically showing the relative movement of two panels in the panel bonding step of Fig. 3. In the bonding step shown in Fig. 3, the two panels are mounted on the upper and lower flat plates in a facing state (step S21), and the alignment mark is read by the identification camera (step S22). In step S21, the upper and lower panels are arranged as close as possible. The distance between the upper and lower panels is, for example, about 2 // m. Next, it is determined whether the alignment marks of the upper and lower panels are within a predetermined range (step S23). That is, it is determined whether the upper and lower panels have been aligned with a predetermined accuracy. If the upper and lower panels have not been aligned with a predetermined accuracy, the lower plate is moved in the X, γ, and θ directions by 1245944 (step S24), and the alignment is performed again. The processing of step S24 is not just to align the lower panel with the upper panel, but to move the lower panel to a position where the amount of misalignment during compression is taken into account. After the alignment based on the predicted position is completed, a pressure of, for example, 300 kg is applied to the upper and lower panels (step S25). The pressurization at this time is continuously performed from 0 to 300 kg. That is, as shown in Fig. 4, the processing of step S24 is performed in the position of pi, so that the lower panel is moved to the position of the panel indicated by the dotted line as shown by the solid arrow. The position of the panel indicated by the dotted line 'is a position where the amount of misalignment that would occur during pressurization 10 has been taken into consideration in advance. Therefore, when the applied pressure is continuously applied to the upper and lower panels and reaches a final applied pressure of 300 kg, that is, when the upper panel reaches the final position pF as shown by the dashed arrow in Fig. 4, the alignment of the upper and lower panels ends. In addition to the above examples, there are methods to simultaneously press the substrate in both the vertical direction and the horizontal direction when pressing two substrates. That is, a method of detecting the misalignment of the panel group while pressing, and correcting the misalignment during pressing based on the detection result. (For example, refer to Japanese Patent Literatures 2 and 3.) [Japanese Patent Literature 1] Japanese Patent Publication No. HM0508 No. 20 [Japanese Patent Literature 2] Japanese Patent Publication No. 2000-25 8746 [Japanese Patent Literature 3] Japanese Patent Laid-Open Publication No. 2002-40398 Rabbit 1245944 The problem of panel bonding shown in the above figure i is that the panel group is relatively pressurized in stages to perform relative bonding while the panel group is relatively moved. Therefore, even if the centers of the two panels are correctly aligned (aligned) in the second stage (the position of P1 in FIG. 2), as long as the position of the second stage (the positions of the hits, the decays, and the two in FIG. 2) ) The center position deviation caused by the day guards must be aligned at each stage. Therefore, when the pressing step and the alignment step are performed alternately, the time for performing the alignment in the panel bonding step increases, and the time required for the panel bonding step increases accordingly. In addition, in the alignment of each stage, the panel group is relatively moved under the state of pressing two panels (for example, 30 kg, 60 kg, 100 kg). At this time, shear stress will be generated on the sealing adhesive between the two panels, which may cause deformation or damage to the sealing adhesive. When the sealing adhesive material is damaged, the liquid crystal that is sealed between the two panels should leak out ', making the liquid crystal display device unusable. Moreover, in the panel bonding step shown in FIG. 3 described above, the two panels are also relatively moved when the pressing step is performed at 10 15. Therefore, the adhesive material may also be damaged. However, the relative movement of the panel at this time is not caused by the alignment operation, but is caused by the error of the pressure mechanism. The method of measuring and correcting the misalignment during pressurization is to correct the misalignment when a certain degree of misalignment is detected. To perform misalignment correction within a short pressurizing time, it is necessary to expand the allowable range of the amount of misalignment to a certain extent and reduce the number of misalignment corrections performed since the start of pressurization and before the end of pressurization. However, when the permissible range in the misalignment is enlarged, the relative movement amount (displacement correction amount) of the panel under pressure increases, which may cause damage to the sealing adhesive. In addition, in order to avoid damage to the sealing adhesive, it is necessary to increase the displacement correction reading, and the time for the pressure step 20 1245944 is increased. [Summary of the Invention] Summary of the Invention This design is a system that can be made in a lossy way, and the purpose is to provide ~ inserted in the panel bonding step, shorten the alignment time, and reduce the sealing material Damaged panel bonding method and device. Means for mountain seals to solve the aforementioned problems 10 15 In order to solve the aforementioned problems, the present invention provides a panel bonding method ^ is to make two pieces of panel pressed together through an adhesive material to form a superimposed shape L, and, In the state where the panel faces of the two panels face, the two panel groups are pressurized with a predetermined external pressure through the adhesive material in a direction perpendicular to the panel surface, and are added after the predetermined Before pressure, move at least one of the two panels that move forward in a direction parallel to the panel surface, which is equivalent to the predicted misalignment amount predicted and set in advance, and when the pressure is reached beyond the predetermined pressure, Pressing ends when the two panels are aligned. According to the panel bonding method of the present invention, as described above, the panel is attached to the panel B, and the lower panel is moved while pressurizing the upper panel, so that the displacement between the two panels is removed at the same percentage as the pressure, so It can reduce the relative movement of the two panels to a minimum. Therefore, it can prevent deformation or damage of the sealing adhesive material installed between the panels, and can achieve high-precision alignment function and high-reliability paste. Closed panel bonding device. The panel bonding method of the present invention can also temporarily stop the aforementioned pressure before starting to press to 1245944, and measure the allowable amount of the panel group before stopping, and when the amount of the dislocation exceeds the predetermined amount of dislocation, at least One of the two panels is moved in a direction parallel to the panel surface to correct the misalignment amount, and the pressure is started again after the misalignment amount correction is completed. 5 Alternatively, the amount of dislocation of the aforementioned panel group may be measured before the pressurization starts to the end, and when the amount of the dislocation exceeds a predetermined amount of dislocation, one side of the two panels is pressurized continuously while at least one of the two panels is pressed. The displacement is corrected in a direction parallel to the panel surface. Also, the aforementioned panel bonding method is to make the measured distance between the aforementioned misalignment 10 and the two panels have a correlation and store it as information 'to set the predicted amount of misalignment in the subsequent panel bonding step. Better. Alternatively, it is also possible to correlate the measured amount of displacement with the amount of pressurization and store it as a lean message to set the predicted amount of displacement in the subsequent panel bonding step. In addition, the panel bonding method of the present invention can issue an alarm to notify the occurrence of an abnormal condition when the amount of misalignment measured above is greater than the value of the misalignment obtained from the misalignment information stored previously. 20 ’The present invention provides a panel bonding device, which is a device for laminating two panels with a bonding material plus I and bonding together. Including the upper T plate, the upper plate moving mechanism, the table, the photographic plate, ^ T flat phase separation material «The two pieces of moving-oriented mechanism are used to make the upper plate move to the vertical side. The table is used to make the aforementioned one, and the camera is used for image recognition of the alignment marks 12 1245944 of the two panels that are respectively moved in the horizontal direction and the two panels, and the aforementioned control device is used to Control the movement of the upper plate moving mechanism, and control the movement of the table to perform the alignment of the panel group by the image recognition result obtained by the scene camera ^, and the control device moves the upper plate on one side The mechanism operates and pressurizes the aforementioned five panels, while controlling the movement of the uranium table so that the movement of the aforementioned lower plate is equivalent to a movement distance predicted and set in advance by a predicted misalignment amount. According to the panel bonding device of the present invention, when the panel is bonded, the lower panel is moved while pressing the upper panel, and the displacement between the two panels is removed while pressing, so that the two panels can be made. The relative movement of the panel 10 is reduced to a minimum. Therefore, it is possible to prevent deformation or damage of the sealing adhesive material provided between the panels, and to improve the reliability of the panel bonding device. In addition, the moving distance of the panel during the alignment operation can be reduced as much as possible, and the time required for the bonding process can be shortened. Therefore, it is possible to improve the performance of the panel bonding device. In the panel bonding device of the present invention, it is preferable that the control device includes a memory device and a prediction mechanism. The measured horizontal position of the aforementioned lower plate ~, and λ are stored, and the mechanism is used by the predictor to store the aforementioned misalignment amount. χ, the aforementioned control device '= structure: the alarm mechanism is used to notify an alarm person that an abnormal condition has occurred when the detected misalignment amount exceeds a predetermined value based on' ::: τ '. . tEmbodiment 3 Detailed Description of the Preferred Embodiment 13 1245944 Next, this embodiment will be described together with the illustration. First, the panel bonding apparatus using the panel bonding method of the present invention will be described with reference to FIG. 5. Fig. 5 is a cross-sectional view of a panel bonding apparatus using the panel bonding method of the present invention. 5 The panel bonding device 10 shown in Fig. 5 is a device for pressing and adhering the TFT substrate (TFT panel) and the color substrate (CF panel) constituting the liquid crystal display device. The TFT panel is formed with a liquid crystal driving circuit, and the information is displayed on the panel by the liquid crystal sealed between the Ding panel and the CF panel. The panel bonding device 10 has an upper plate 14 and a lower plate 18, and an upper plate 14 10 series are installed with CF panel 2 and can be moved in the vertical direction (Z-axis direction), while the lower flat plate 18 is installed with TFT panel 16 and can be moved in the horizontal direction. The upper and lower flat plates 14 and 18 are arranged by the upper chamber 20 and The chamber formed by the lower chamber 22. When the upper chamber 20 and the lower chamber 22 are sealed, the chamber can be in a vacuum state, and the chamber can be set to a predetermined vacuum pressure. 15 The upper chamber 20 is connected to the support member 24, and the support member 24 is a gas cylinder, etc. The reciprocating mechanism is configured to move in the vertical direction (two-axis direction). The opening relationship of the chamber is performed by moving the upper chamber 20 up and down. The upper plate 14 is connected to the support member 26 that extends through the wall of the upper chamber 20 and extends to the interior. By moving the support member 26 in the vertical direction 20 (Z-axis direction) by the upper plate moving mechanism 28, the indoor upper plate 14 can be moved in the vertical direction (Z-axis direction). In addition, the 'upper plate moving mechanism 28 can A well-known linear motion mechanism, such as a ball screw mechanism or a linear motor, is shown in the figure. In the example shown in FIG. 5, the servo motor 30 for driving the ball screw is used as a driving source to drive the upper flat 1245944 plate moving mechanism 28 so that The upper plate 14 moves in the vertical direction (z-axis direction). The lower plate 18 is disposed in the lower chamber 22, and is connected to a supporting member 32 extending through the bottom of the lower chamber 22. The supporting member 32 is fixed to the χγ0 table 34 and the lower plate. 18 can be moved horizontally in the room. That is, by moving 5 sets of χγθ tables 34 in the horizontal plane in the X direction, υ direction, and θ direction (rotation direction), the lower plate can be moved indoors in the X direction, γ direction, And the θ direction (rotation direction). Since the XYZ table 34 including the driving source can be constituted by a well-known moving mechanism, detailed descriptions thereof are omitted. In the aforementioned panel bonding device, the upper plate moving mechanism 28 is installed in a solid state. A frame 36, and a guide mechanism 38 for guiding the upper plate moving mechanism 28 to move in the vertical direction are mounted on the frame 36. The guide mechanism 38 guides the upper plate moving mechanism 28 so that the upper plate is flat. 14 A mechanism that can move vertically with good accuracy, but due to mechanical size errors or installation errors, it will not be a completely vertical direction. 15 That is, the vertical movement of the upper flat plate 14 relative to the lower flat plate 18 is not completely vertical. When the upper plate 14 moves in the vertical direction, it will deviate slightly from the lower plate 18 in the horizontal direction (X, Υ, 0 direction). In order to correct such a deviation, the lower plate 18 is fixed to the XΥΘ table 34 and can be horizontal Move in the direction. That is, by driving the XYZ table 34, the alignment of the upper and lower panels when they are mounted on the upper and lower plates 14, 20 and 18, and the correction of the misalignment generated when the pressure is applied. The upper plate moving mechanism 28 is provided for the alignment. The image recognition camera 40 as a sensor at the time, the image recognition camera 40 aligns the CF panel 12 and the TFT panel 16 respectively through a transparent window provided in the upper chamber 20 and a through opening provided in the upper plate 14. Mark for image recognition. That is, — 15 1245944 uses the image recognition camera 40 to perform image recognition on the alignment marks of the panels 12 and 16 on both sides, and aligns the CF panel and the TFT panel. In addition, in order to recognize the deviation of the alignment mark in the X, Y, and 0 directions, the image recognition camera 40 is preferably set at two or more different positions. The upper flat plate moving mechanism 28 is provided with a load sensing line 42 as a load sensor. The load sensor 42 can detect the load (ie, the external pressure) of the CF panel 12 mounted on the upper flat plate 14 when pressed against the TFT panel 16 mounted on the lower flat plate 18. Among them, the operation of the servo motor 10 30 as a driving source for vertically moving the upper plate 14 is controlled by the control device 44. In addition, the χγθ table 34 for causing the lower plate to move privately in the horizontal direction is also controlled by the control device. The control device 44 receives the alignment signal displayed by the image recognition result of the image recognition camera 40, and the external pressure lean signal displayed by the detection result of the external pressure of the load sensor 42. Therefore, the control device 44 controls the operations of the upper plate moving mechanism 28 and the XYZ table 34 based on the alignment information from the aforementioned image recognition 15 camera 40 and the pressure information from the load sensor 42. In addition, in order to learn the alignment and misalignment correction as described later, the control device 44 should have a memory (memory mechanism) for storing various information, and a processor (such as a misalignment correction amount and a correction amount for nose misalignment based on the prediction). Forecasting agency). 20 Also, the control device 44 should have a warning device (alarm mechanism), and the warning device can determine that the device has an abnormal condition and send out an alarm when the difference between the measured amount of displacement and the actual amount of displacement exceeds a predetermined value. . In addition, in the example shown in FIG. 5, the panel 12 ′ is installed on the upper plate 14 and the TF panel 16 is installed on the lower plate 18. However, the TFT panel 16 may be installed on the upper plate 12 1245944 and A CF panel 12 is mounted on the lower plate 18. Hereinafter, a panel mounted on the upper plate 14 is referred to as an upper panel, and a panel mounted on the lower plate µ is referred to as a lower panel. In addition, when the TFT panel and the C! F panel are bonded, a sealing adhesive material is provided in a frame shape at five portions around the lower panel, and liquid crystal is supplied to the inside thereof. Liquid crystal is sealed between the upper and lower panels by pressing and adhering the upper and lower panels with a sealing adhesive. For example, a UV-curing adhesive can be used, and it can be completely cured by irradiating ultraviolet rays after the lamination process as a sealing adhesive. Next, the alignment method of the first embodiment of the present invention will be described with reference to Figs. 6 to 8. FIG. 6 is a flowchart showing a panel bonding step using the panel bonding method according to the first embodiment of the present invention. 7 and 8 are diagrams schematically showing the relative movement of the two panels in the panel bonding step shown in FIG. 6. The panel bonding step shown in FIG. 6 uses the panel 15 bonding device 10 shown in FIG. 5. First, the upper and lower panels are mounted on the upper and lower flat plates 14 and 18 in an opposing state (step S31), and The image recognition camera 40 reads the alignment marks set on the upper and lower panels 12 and 16 (step S32). Next, it is determined whether the alignment marks of the upper and lower panels are within a predetermined range (step S33). That is, it is determined whether the upper and lower panels have been aligned with a predetermined accuracy of 20 °. When the alignment has not been performed with a predetermined accuracy, the lower plate 18 is moved in the directions of γ, = (step S34), and alignment is performed again. The above steps are steps of rough alignment (rough alignment). Rough alignment step_ Aligns the alignment of the upper and lower panels to, for example, 5 // m or less. After the rough alignment is completed, perform fine alignment (fine alignment). The fine alignment of 17 to 245944 is the same as that of steps S32 to S34 described above, and the alignment is performed so that the displacement of the upper and lower panels is, for example, 0.5 // m or less. At the end of the fine alignment, the upper plate 14 is moved vertically with respect to the lower plate 16 and pressurized in a state where the two panels are overlapped. At this time, when the final pressure 5 is set to 1001 ^, the pressure is first increased to, for example, 30 kg (step S38). As described later, this step is performed when the upper plate 14 is moved vertically to press the upper plate to the lower plate, and the lower plate 18 (lower plate) is moved to follow the displacement of the upper plate. That is, the horizontal movement of the upper panel is predicted, and the lower panel is also moved in the same direction at the same time (predicted movement). After pressing, the alignment marks of the upper and lower panels are read (step 39), and it is determined whether the misalignment of the upper and lower panels is within the allowable range (step S40). When the misalignment of the upper and lower panels exceeds the allowable range, the XYZ table 34 is driven to move the lower plate 18 (lower plate), and the fine alignment of the upper and lower panels is performed again. Then, the process returns to step S38, and the processes of steps S39 and S40 are performed. When it is determined in step S40 that the misalignment of the upper and lower panels is within the allowable range, it is determined whether the pressure previously applied is finally increased (step 42). If it is not the final pressurization, the process returns to step S38, and steps S38 to 2 are performed again. In the example shown in Fig. 6, the final pressurization is made §, and during the pressurization, fine alignment is performed at the positions of 30 kg and 60 kg. That is, the processing in the first step 20 to step S38 is first pressurized to the withdrawal g, and the processing in the next step s38 is pressurized to 60 kg, and the processing in the next step S38 is pressurized to the dirty §. When the pressure is increased to 100 kg, the pressure is determined to be the final pressure. Figures 7 and 8 are diagrams schematically showing the movement of the upper and lower panels in the aforementioned panel bonding step. Figure 7 shows an example of the _movement of the lower panel 1245944 corresponding to the misalignment of the upper panel. In addition, Fig. 8 shows an example in which the displacement of the lower panel relative to the upper panel is predicted to move. However, since an error occurs, only a few corrections can be made during the pressurization. In Fig. 7 (a), the position of the upper panel in the z-axis 5 direction is the position of P2 under the pressure of 30 kg. When the upper panel is moved from pi to P2 in the vertical direction, as shown by the dotted arrow, a misalignment occurs in the X and Y directions. This misalignment is mainly caused by a mechanical error for moving the plate moving mechanism 28 above the upper plate 14. In order not to cause the aforementioned misalignment, when the upper panel is moved from P1 to P2, 'as shown in Fig. 7 (b), the lower panel is moved at the same time as the upper panel is displaced by 10 square water (the processing of the initial step S38) . In the example shown in Fig. 7, in the position of P2, it is judged that the dislocation of the panel is within the allowable range (YES in step S40). Therefore, the process returns to S38, and 60 kg of pressure is applied to the upper and lower panels. When the applied pressure reaches 60 °, the position of the upper panel is at P3. When the upper panel is moved from P2 to P3, 15 misalignments also occur. Therefore, when the upper panel is moved from M to the position of P3 (that is, when the applied pressure is increased from 30kg to 60kg, the lower panel is also predicted to move (the processing of the second step S38). Next, the final pressure i100kg is applied to the upper and lower panels. When the applied pressure reaches 100 kg, the position of the upper panel is PF. When the upper panel is moved from P3 20 to the position of PF, the position is also misaligned. Therefore, the applied pressure is increased from 60 kg to At the same time, the lower panel is also predicted to move (the process of the third step of S38). After repeated operations such as the above pressurization, when the pressurization before the determination in step S42 is the final pressurization, the alignment marks of the upper and lower panels are measured. The amount of misalignment (step 19 1245944 S43), the processing of the panel is completed. For example, if the panel ™ ~ one or two pieces are based on the above t, and the lower panel is oriented horizontally = it will not occur when the upper panel is moved horizontally. Production 4 Therefore, the panel is only misplaced even if it is pressurized, and the sealing material is only misplaced. Therefore, it is sandwiched between the upper and lower panels to be pressed: the material is pressed in a direction other than the straight direction, which can prevent the product from being sealed to the adhesive material. Shearing in horizontal direction Force to damage it. 10 15
傷，::晶::止!面板貼合時之錯位造成密封黏著材損 曰曰。之情況’而可提升貼合步驟之可靠性。又， 階段性之城途中不需要細微貼合，因此，可減少如此之 步驟所需之相，而可縮短貼合步驟所需之時間。 另外第8圖所示之例中，係於上面板垂直移動之同時 使下面板朝7jc平方向則彳移動u於加壓之各階段中， 仍會產生應修正之辑M , lL ^ y ^ Μ— 之錯位。如此之例，係例如上面板之錯位 與事先設定之下面板之預測移動量（即，上面板之錯位之預 測值)不同之情況，相當於貼合裝置1()之移動機構之精度經Injury, :: 晶 :: 止! Dislocation of the panel when it is bonded may cause damage to the sealing material. In this case ', the reliability of the bonding step can be improved. In addition, fine bonding is not required during the staged city. Therefore, the number of phases required for such a step can be reduced, and the time required for the bonding step can be shortened. In addition, in the example shown in FIG. 8, when the upper panel is vertically moved and the lower panel is moved toward the 7jc horizontal direction, the u is moved in the various stages of pressure, and the series M, lL ^ y ^ that should be corrected will still be generated. Μ— Dislocation. Such an example is the case where the misalignment of the upper panel is different from the preset movement amount of the lower panel (that is, the predicted value of the misalignment of the upper panel), which is equivalent to the accuracy of the movement mechanism of the bonding device 1 ().
時地變化之情況。 ί 於第8(a)圖中，加壓至扣以之狀態下上面板於z軸方向 之位置係P2之位置。事先預測上面板於垂直方向上由η移 20動至Ρ2之間產生之水平方向之錯位係以虛線之箭號表示。 其中，當上面板之移動偏離預測值，且實際上係如實線之 箭號所示般移動時，於Ρ2位置中產生些許錯位。因此，於 Ρ2位置中修正該錯位。該修正相當於第6圖之步驟S41中之 修正。 20 1245944 因此，於正施加例如30kg之外加壓力之狀態下，移動 I面板進行位置修正。然而，由於τ面板已事先進行預測 移動，因此如第8(b)圖之實、線之箭號所*，下面板之移動量 5僅是偏離預測移動位置之錯位量，可使上下面板於加廢下 之相對移動量（即，Ρ2中之下面板之移動量)極小。即，可使 中下面板之移動量之值遠小於藉習知方法進行之中下 面板之移動量之值。 第8圖所示之例中，於加壓至60kg之位置Ρ3及最終加壓 至l〇〇kg之位置PF中，亦與於位置？1之位置修正相同地進行 10 偏離預測位置之修正。 如前述，第6圖所示之貼合步驟，於加壓之各階段判定 偏離預測位置之錯位。且，於各階段中，當偏離在容許範 圍内時，上下面板係如第7圖所示般移動，另外，於各階段 中，當偏離超出容許範圍時，上下面板係如第8圖所示般移 15 動。 接著，針對本發明之第2實施例，一面參照第9圖至第 14圖一面說明。第9圖係顯示本發明之第2實施例之面板貼 合步驟之一部份的流程圖。 本發明之第2實施例之面板貼合步驟，係於第6圖所示 20之貼合步驟中連續地進行步驟S38〜步驟S42之處理。粗略對 位及細微對位之處理，都與第6圖所示之步驟S31〜S37之處 理相同。 即，細微對位結束時，如第9圖所示，首先開始加壓z 軸方向之加壓（步驟S51)。然後，於加壓中利用影像辨識攝 21 J245944 影機補解標騎行影像_(步驟S52)，且，取得上面 板目前於z軸方向上之位置（步驟853)。2軸方向位置，只要 測出上平板Η於Z軸方向上之位置即可。又由於由負荷感 知器42測出之載重（外加壓力）係大致與冰方向之位置成 5正比例，因此’亦Τ將該外加壓力視作冰方向之位置。 接著’根據步驟S52之影像辨識結果，演算上面板於取 得之Ζ軸位置中之水平方向上之位置(χ、γ、0之位置)(步 驟糾。且’根據步驟S54之演算結果，演算上下面板間之 實際之錯位量（步驟S55)。 10 錢’判定於步鄉55求出之錯位量是不是於容許值内 (步驟S56)。此時，錯位量之演算係分別針對χ/γ/θ各方向進 行。/!定錯位量超出容許值時，為了修正其錯位量，演算 桌檯應該移動之移動量（步驟S57)。 之後，將於步驟s57求出之桌檯之移動量通知ΧΥΘ卓 15檯34(步驟S58)，使ΧΥΘ桌㈣移動於步驟S57求出之桌檯 之移動量（步驟S59)。 ’、 藉前述步驟，於實際之錯位量與預測之錯位量之差超 出容許值時，進行修正上下面板之錯位量之處理。 —進行錯位修正後，本實_係取得已藉前述步驟奶 2。演算之錯位量作為資訊加以儲存(步驟_。然後，根據儲 存之資訊，作成並變更用以作為演算錯位量之預測值根據 之X/Υ/θ方向之錯位量與2軸方向之位置之關係（參照第⑺ 圖至第12圖）。即’學習實際之錯位量與預測之錯位量之 差’亚於之後之貼合步驟中，學習實際之錯位量之變化以 22 1245944 使錯位量之預測值更加正確。 ，前述之步驟S56中，肖定實際之錯位量與預測之錯位量 ^於容許值内時，處理進入步驟_。即，亦儲存有關= —内之錯位量之資訊，並於之後之貼合步驟中，學習 貫際之錯位量之變化以使錯位量之預測值更加正確。此 外，本實施例中，錯位量之學習係分別針對χ/Υ/θ方向個 10 錯位量之學f結束後，處理進人步驟S42(參照第6 圖）’並判定是不是已到達預定之最終加壓。 以上之步驟’係-面使上面板朝z軸方向移動（即 增加外加壓力）—面進行，因此可於進行加壓步驟之同 订錯位修正’而貼合步驟所需之時間僅是加髮步驟所 時間°即4於修正錯位時必需之下面板之移動係於力而 步驟中進仃，因此，可縮短面板貼合步驟所需之時間。Changing situation. ί In Fig. 8 (a), the position of the upper panel in the z-axis direction under the state of being pressed is the position of P2. It is predicted in advance that the horizontal displacement caused by moving the upper panel from n to 20 in the vertical direction to P2 is indicated by a dashed arrow. Among them, when the movement of the upper panel deviates from the predicted value, and actually moves as shown by the solid arrow, some misalignment occurs in the position P2. Therefore, the misalignment is corrected in the P2 position. This correction corresponds to the correction in step S41 of FIG. 20 1245944 Therefore, while applying pressure other than 30kg, move the I panel to correct the position. However, since the τ panel has been predicted to move in advance, as shown by the solid and line arrows in Figure 8 (b) *, the amount of movement of the lower panel 5 is only the amount of misalignment from the predicted movement position, allowing the upper and lower panels to The relative moving amount under the waste (ie, the moving amount of the lower panel in P2) is extremely small. That is, the value of the amount of movement of the middle and lower panels can be made much smaller than the value of the amount of movement of the middle and lower panels by conventional methods. In the example shown in Fig. 8, in the position P3 pressurized to 60 kg and finally to the position PF pressurized to 100 kg, is it also related to the position? The position correction of 1 similarly performs the correction of 10 deviations from the predicted position. As described above, in the bonding step shown in Fig. 6, the displacement from the predicted position is determined at each stage of the pressurization. And, in each stage, when the deviation is within the allowable range, the upper and lower panels move as shown in Fig. 7, and in each stage, when the deviation exceeds the allowable range, the upper and lower panels are as shown in Fig. 8. 15 moves in general. Next, a second embodiment of the present invention will be described with reference to Figs. 9 to 14. Fig. 9 is a flowchart showing a part of the panel attaching steps of the second embodiment of the present invention. The panel bonding step of the second embodiment of the present invention is performed continuously in steps S38 to S42 in the bonding step of 20 shown in Fig. 6. The processing of rough alignment and fine alignment is the same as that of steps S31 to S37 shown in FIG. That is, when the fine alignment ends, as shown in FIG. 9, the pressure in the z-axis direction is first started (step S51). Then, during the pressurization, the image recognition camera 21 is used to supplement the riding image (step S52), and the current position of the top panel in the z-axis direction is obtained (step 853). For the position in the 2-axis direction, it is only necessary to measure the position of the upper flat plate in the Z-axis direction. Since the load (applied pressure) measured by the load sensor 42 is approximately proportional to the position in the ice direction, ′ also regards the applied pressure as the position in the ice direction. Then 'calculate the position of the upper panel in the horizontal direction (the position of χ, γ, 0) among the obtained Z-axis positions according to the image recognition result of step S52 (step correction.) And calculate the upper and lower positions according to the calculation result of step S54. The actual amount of misalignment between the panels (step S55). 10 Money 'determines whether the amount of misalignment obtained at Buxiang 55 is within the allowable value (step S56). At this time, the calculation of the amount of misalignment is for χ / γ / θ is performed in all directions. /! When the amount of misalignment exceeds the allowable value, in order to correct the amount of misalignment, the amount of movement of the table should be calculated (step S57). After that, the amount of movement of the table obtained in step s57 will be notified χΥΘ 15 sets of 34 (step S58), move the XY table ㈣ to the moving amount of the table (step S59) obtained in step S57. ', With the foregoing steps, the difference between the actual misalignment amount and the predicted misalignment amount exceeds the allowable When the value is changed, the process of correcting the misalignment amount of the upper and lower panels is performed.-After performing the misalignment correction, the real _ is obtained by borrowing the previous step 2. The calculated misalignment amount is stored as information (step_. Then, based on the stored information, Make and change The relationship between the amount of misalignment in the X / Υ / θ direction and the position in the 2 axis direction used as the predicted value for calculating the amount of misalignment (refer to Figures 至 to 12). That is, 'learn the actual amount of misalignment and the predicted misalignment. In the following step, the difference between the actual amount of displacement and the actual amount of displacement is adjusted to 22 1245944 to make the predicted value of the displacement more correct. In the foregoing step S56, the actual displacement and the predicted displacement are determined. ^ When it is within the allowable value, the process proceeds to step _. That is, the information on the amount of misalignment in = — is also stored, and in the subsequent bonding step, the change in the amount of misalignment is learned to make the predicted value of the amount of misalignment. It is more correct. In addition, in this embodiment, the learning of the amount of misalignment is performed for each of the 10 amounts of misalignment in the χ / ， / θ direction, and the process proceeds to step S42 (refer to FIG. 6) to determine whether it has been The predetermined final pressurization is reached. The above steps are 'moving the upper panel in the z-axis direction (ie, increasing the applied pressure)-proceeding on the surface, so it can be applied in the same way as the step of correcting the misalignment of the pressurizing step. The time required is only to add The time required for the step is 4, which means that the movement of the lower panel must be performed by force while correcting the misalignment. Therefore, the time required for the panel bonding step can be shortened.
又，因為係學習預測錯位量與實際之錯位 變更預測錯位量使其更接近實際錯位量之值， 量之差， 而 因此， 可減 20 少上下面板間之相對移動量。藉此，可防止夹置於面 之密封黏著材之損傷，而達成可靠性高之面板貼合處理反間 第13圖及第14圖係模式地顯示第9圖所示之貼合心 中上面板與下面板移動之圖。第13圖係錯位量之預測值^ 貫際之錯位篁-致之情況之例’第14圖則係錯位量之預^ 值與實際之錯位量之間產生誤差而進行了修正 、列In addition, because the predicted misalignment amount and the actual misalignment are changed, the predicted misalignment amount is changed to be closer to the actual misalignment amount and the difference between the amounts. Therefore, the relative movement amount between the upper and lower panels can be reduced by 20%. In this way, damage to the sealing adhesive material sandwiched on the surface can be prevented, and a highly reliable panel bonding process can be achieved. FIGS. 13 and 14 schematically show the upper center panel of the bonding center shown in FIG. 9. Figure with the bottom panel moving. Fig. 13 is the predicted value of the amount of misalignment ^ Example of the situation where the misalignment is consistent. Fig. 14 is the correction of the amount of misalignment between the amount of misplacement and the actual amount of misalignment.
例 之情況之 因此，第9圖所示之貼合步驟中，當錯位量之預測值與 23 1245944 實際之：位量一致時’上面板如第i3⑷圖之實線箭號所 丁由取先對位之位置P1移動至最終位置砰時，下面板會 如第13(b)圖之實線箭號所示，移動該上面板於水平方向I 移動之量。因此，前述上下面板於水平方向上並非相對移 5動’由Z轴方向觀察時，上下面板看起來是於重疊狀態下^ 樣作水平移動。因此，夹置於上下面板間之密封黏著材， 僅受到外加壓力施壓，沒有剪切應力。 另外，第9圖所示之貼合步驟中，當錯位量之預測值斑 實際之錯位量之間產生誤差而已進行錯位量之修正時，1 1〇面板之移動，係與第14⑷圖之虛線所示之預測路徑不同， 實際上係實線箭號所示之路徑。因此，下面板之移動不是 第14⑻圖所示之虛線箭號之路徑，而是如實線箭號所示般 已進行位置修正之路徑。 然而’由於偏離預測錯位量之誤差藉學習而為極小之 Η值，因此，錯位修正之判斷基準之容許值可為小值，錯位 量超出容許值之可能性或次數因而少。因此，如第14圖所 不，即使進灯修正’上下面板間之相對移動量仍小，不至 於會損傷上下面板間之密封黏著材。 第15圖係顯tf第9圖中進行面板貼合步驟時，控制裝置 20 内各種資訊之流程之圖。 控制裝置44中，影像辨識部5〇由影像辨識攝影機40得 到對準標諸之讀取結果時，由影像辨識攝影機4〇之讀取結 果演算X方向之錯位量财方向之錯位量，並送至χγθ偏離 算部52。ΧΥΘ偏離算部52係根據χ方向之錯位量訂方向之 24 1245944 錯位里决疋χυθ桌檯34之移動量。即，演算χγθ桌棱μ於 Χ、Υ、θ方向上之錯位修正量（桌檯之ΧΥΘ值）。 ΧΥΘ桌檯34之錯位修正量，係送至學習反映部54，且 將已學白之錯位變更成補充後之錯位修正量（ΧΥΘ值）。接 5著，將經過補充之學習之錯位的錯位修正 量送至比較、偏 離//、^rrcr卩56比較、偏離演算部56係比較經過補充之學習 之錯位的錯位修正量，與同-Z軸位置中錯位修正量之預測 值，而決疋该Z值位置中之錯位修正量（χγθ/ζ修正量）。 比較、偏離演算部56所決定之錯位修正量，於送至驅 10動控制部58之同時亦送至ΧΥ0趨勢學習部60。驅動控制部 58根據接收之錯位修正量控制χγθ桌檯34之移動，而進行 錯位修正。又，χΥΘ趨勢學習部6〇係根據接收之錯位修正 量求出錯位量之趨勢，並供給至χγ0目標位置演算部62。 ΧΥΘ目彳示位置演异部62係一面反映錯位量之趨勢，一 15面演算錯位量之預測值(ΧΥΘ之目標值），並如前述供給至比 較、偏離演算部56。 又，驅動控制部58由負荷感知器42或其他位置檢測器 接收上平板18之有關Ζ軸位置之資訊，且驅動控制部咒將乙 軸位置之值(Ζ值)供給至學習反應部5 4及X Υ Θ目標位置演算 20 部 62。 且，藉將比較、偏離演算部56所決定之錯位修正量亦 送至警報產生部64，可於錯位修正量超出預定之值時，判 斷裝置發生異常現象，且發出警報通知裝置之操作員。即， 警報產生部64可於比較、偏離演算部所決定之錯位修正 25 1245944 量，即上下面板之錯位量超出預定之值時，發出警報。 由於一次之貼黏步驟與下一次之貼黏步驟係不太有經 時變化，因此預測錯位量與實際之錯位量之差原本就是接 近零之極小之值。故，監視以比較、偏離演算部56決定之 5 錯位修正量，可於其值大於預定值時，判斷裝置發生了異 常狀況。此時，藉發出警報，將前述異常狀況通知操作員， 可提昇裝置之安全性。 此外，如前述之結構之控制裝置44，可藉由處理器及 儲存器等構成之週知硬體結構條列預定之程式來實現。 10 如以上之說明，依據本發明進行面板之貼合時，由於 係一面加壓上面板一面使下面板移動，除去兩片面板間伴 隨加壓產生之錯位，因此可使兩片面板之相對移動量為最 小限度。因此，可防止設置於面板間之密封黏著材之變形 或損傷。所以，可防止封入面板間之液晶等之内容物漏出， 15 而提昇面板貼合裝置之可靠性。 又，可使對位動作中面板之移動距離極小，而可使貼 合處理必需之時間縮短。因此，可提升面板貼合裝置之性 能。 且，當兩片面板中心偏離預定之距離以上時，判定異 20 常並停止裝置，且發出警報，因此，面板貼合裝置之安全 性提昇。 L圖式簡單說明3 第1圖係顯示習知面板貼合步驟之一例之流程圖。 第2圖係模式地顯示第1圖之面板貼合步驟中兩片面板 26 1245944 相對移動之圖。 第3圖係顯示使用事先使面板移動預測錯位量之方法 之面板貼合步驟之一例之流程圖。 第4圖係模式地顯示第3圖之面板貼合步驟中兩片面板 5 相對移動之圖。 第5圖係使用本發明之面板貼合方法之面板貼合裝置 之截面圖。 第6圖係顯示使用本發明之第1實施例之面板貼合方法 之面板貼合步驟之流程圖。 10 第7(a)、（b)圖係模式地顯示第6圖之面板貼合步驟中兩 片面板相對移動之圖。 第8(a)、（b)圖係模式地顯示第6圖之面板貼合步驟中兩 片面板相對移動之圖。 第9圖係顯示本發明之第2實施例之面板貼合步驟之一 15 部分之流程圖。 第10圖係顯示X方向上之錯位量與Z軸方向之位置之 關係之圖表。 第11圖係顯示Y方向上之錯位量與Z軸方向之位置之 關係之圖表。 20 第12圖係顯示Θ方向上之錯位量與Z軸方向之位置之關 係之圖表。 第13(a)、（b)圖係模式地顯示第9圖所示之貼合步驟中 上面板與下面板之移動之圖。 第14(a)、（b)圖係模式地顯示第9圖所示之貼合步驟中 27 1245944 上面板與下面板之移動之圖。 第15圖係顯示進行第9圖所示之面板貼合步驟時控制 裝置中各種資訊之流程之圖。 【圖式之主要元件代表符號表】 10...面板貼合裝置 38...導引機構 12...CF 面板 40...影像辨識攝影機 14...上平板 42...負荷感知器 16...TFT 面板 44…控制裝置 18...下平板 50...影像辨識部 20...上部室 52...ΧΥΘ偏離算部 22...下部室 54…學習反映部 24,26,32...支持構件 56...比較、偏離演算部 28...上平板移動機構 58...驅動控制部 30…伺服馬達 60...ΧΥΘ趨勢學習部 34…ΧΥΘ桌檯 62...ΧΥΘ目標位置演算部 36…框架 64...警報產生部Therefore, in the fitting step shown in Fig. 9, when the predicted value of the displacement amount is the same as the actual value of 23 1245944: when the displacement amount is the same, the top panel is taken as the solid line arrow in Fig. I3. When the aligned position P1 is moved to the final position, the lower panel will move the upper panel in the horizontal direction I as shown by the solid arrow in Figure 13 (b). Therefore, the upper and lower panels do not move relative to each other in the horizontal direction. When viewed from the Z-axis direction, the upper and lower panels appear to move horizontally in an overlapping state. Therefore, the sealing adhesive material sandwiched between the upper and lower panels is only pressed by the external pressure, and there is no shear stress. In addition, in the bonding step shown in FIG. 9, when an error occurs between the predicted value of the displacement amount and the actual displacement amount, and the displacement amount has been corrected, the movement of the 1 10 panel is the same as the dotted line in FIG. 14 The predicted paths shown are different and are actually the paths shown by the solid arrows. Therefore, the movement of the lower panel is not the path of the dashed arrow shown in Figure 14⑻, but the path of which the position has been corrected as shown by the solid arrow. However, because the error from the predicted misalignment amount is extremely small through learning, the allowable value of the judgment criterion for the misalignment correction can be a small value, and the possibility or number of times that the misalignment amount exceeds the allowable value is small. Therefore, as shown in FIG. 14, even if the relative movement amount between the upper and lower panels of the lamp correction is still small, the sealing adhesive material between the upper and lower panels will not be damaged. Fig. 15 is a diagram showing the flow of various information in the control device 20 when the panel bonding step is performed in Fig. 9 in tf. In the control device 44, when the image recognition unit 50 obtains the reading result of the alignment marks from the image recognition camera 40, the reading result of the image recognition camera 40 calculates the amount of misalignment in the X direction and sends the amount of misalignment. To χγθ deviate from the calculation unit 52. The χΥΘ deviation calculation unit 52 determines the movement amount of the 疋 χυθ table 34 in the misalignment according to the amount of misalignment in the χ direction. That is, the amount of correction of the displacement of the χγθ table edge μ in the X, Y, and θ directions (the value of the table XYθ) is calculated. The misalignment correction amount of the XYZΘ table 34 is sent to the learning reflection unit 54 and the learned misalignment is changed to a supplemental misalignment correction amount (XYZΘ value). Next, the supplemental learning misalignment correction amount is sent to the comparison, deviation //, ^ rrcr 卩 56 comparison, and the deviation calculation unit 56 compares the supplementary learning misalignment correction amount, which is the same as -Z. The predicted value of the misalignment correction amount in the axis position depends on the misalignment correction amount (χγθ / ζ correction amount) in the Z value position. The misalignment correction amount determined by the comparison and deviation calculation section 56 is sent to the drive control section 58 and also to the XY0 trend learning section 60. The drive control unit 58 controls the movement of the χγθ table 34 based on the received misalignment correction amount, and performs misalignment correction. The χΥΘ trend learning unit 60 calculates the trend of the amount of errors based on the received misalignment correction amount, and supplies the trend to the χγ0 target position calculation unit 62. The χ 异 Θ shows that the position difference part 62 reflects the trend of the amount of misalignment, and calculates the predicted value of the amount of misalignment (the target value of χΥΘ) on the 15 side, and supplies it to the comparison and deviation calculation unit 56 as described above. In addition, the drive control unit 58 receives the information about the position of the Z axis of the upper plate 18 by the load sensor 42 or another position detector, and the drive control unit supplies the value of the position of the B axis (Z value) to the learning response unit 5 4 And X Υ Θ target position calculation 20 steps 62. In addition, by sending the misalignment correction amount determined by the comparison and deviation calculation unit 56 to the alarm generating unit 64, it is possible to judge that an abnormality occurs in the device when the misalignment correction amount exceeds a predetermined value, and issue an alarm to notify the operator of the device. That is, the alarm generating unit 64 may issue an alarm when the amount of misalignment correction 25 1245944 determined by the comparison and deviation calculation unit, that is, the amount of misalignment of the upper and lower panels exceeds a predetermined value. Since the one step of sticking and the next step of sticking do not change over time, the difference between the predicted misalignment amount and the actual misalignment amount is originally a very small value close to zero. Therefore, the 5 misalignment correction amount determined by the comparison and deviation calculation unit 56 is monitored, and when the value is larger than a predetermined value, it can be judged that an abnormal condition has occurred in the device. At this time, by issuing an alarm and notifying the operator of the foregoing abnormal condition, the safety of the device can be improved. In addition, the control device 44 having the above-mentioned structure can be realized by a predetermined program of a well-known hardware structure composed of a processor, a memory, and the like. 10 As explained above, when the panels are bonded according to the present invention, the lower panel is moved while pressing the upper panel while removing the displacement caused by the pressure between the two panels, so that the two panels can be moved relatively. The amount is minimal. Therefore, it is possible to prevent deformation or damage of the sealing adhesive material provided between the panels. Therefore, it is possible to prevent the contents of the liquid crystal and the like sealed between the panels from leaking out, thereby improving the reliability of the panel bonding device. In addition, the moving distance of the panel during the alignment operation can be extremely small, and the time required for the bonding process can be shortened. Therefore, the performance of the panel bonding device can be improved. In addition, when the center of the two panels deviates by more than a predetermined distance, it is determined that the device is abnormal and the device is stopped, and an alarm is issued. Therefore, the safety of the panel bonding device is improved. Brief description of L-scheme 3 Fig. 1 is a flowchart showing an example of a conventional panel bonding procedure. FIG. 2 is a diagram schematically showing the relative movement of two panels 26 1245944 in the panel bonding step of FIG. 1. Fig. 3 is a flowchart showing an example of a panel bonding procedure using a method of predicting the amount of misalignment by moving the panel in advance. FIG. 4 is a diagram schematically showing the relative movement of the two panels 5 in the panel bonding step of FIG. 3. Fig. 5 is a sectional view of a panel bonding apparatus using the panel bonding method of the present invention. Fig. 6 is a flowchart showing a panel bonding procedure using the panel bonding method according to the first embodiment of the present invention. 10 Figures 7 (a) and (b) are diagrams schematically showing the relative movement of two panels in the panel bonding step of Figure 6. Figures 8 (a) and (b) are diagrams schematically showing the relative movement of two panels in the panel bonding step of Figure 6. Fig. 9 is a flowchart showing part 15 of one of the panel bonding steps of the second embodiment of the present invention. Fig. 10 is a graph showing the relationship between the amount of misalignment in the X direction and the position in the Z axis direction. Fig. 11 is a graph showing the relationship between the amount of misalignment in the Y direction and the position in the Z axis direction. 20 Figure 12 is a graph showing the relationship between the amount of misalignment in the Θ direction and the position in the Z axis direction. Figures 13 (a) and (b) are diagrams schematically showing the movement of the upper panel and the lower panel in the bonding step shown in FIG. Figures 14 (a) and (b) are diagrams schematically showing the movement of the upper panel and the lower panel in the bonding step shown in FIG. Fig. 15 is a diagram showing the flow of various information in the control device when the panel bonding step shown in Fig. 9 is performed. [Representative symbol table of main components of the drawing] 10 ... Panel bonding device 38 ... Guide mechanism 12 ... CF panel 40 ... Image recognition camera 14 ... Upper plate 42 ... Load sensing Device 16 ... TFT panel 44 ... control device 18 ... lower plate 50 ... image recognition unit 20 ... upper chamber 52 ... χΘ deflection calculation unit 22 ... lower chamber 54 ... learning reflection unit 24 , 26, 32 ... Supporting members 56 ... Comparison and deviation calculation section 28 ... Upper plate moving mechanism 58 ... Drive control section 30 ... Servo motor 60 ... XΥΘ Trend learning section 34 ... XΥΘ Table 62 ... XΥΘ target position calculation unit 36 ... frame 64 ... alarm generation unit
2828