1301293 (1) _ 九、發明說明 【發明.所屬之技術領域】 本發明係關於薄膜材料的結晶化方法及其裝置，特別 是關於，藉由對薄膜狀的矽製材料照射脈衝狀雷射，來使 材料結晶化，主要作爲平面面板所使用的薄膜電晶體的結 晶化矽之製造所使用的薄膜材料的結晶化方法及其裝置。 B 【先前技術】 於專利文獻1、2中記載有：於薄膜狀的矽製材料照 射複數次脈衝狀雷射來形成結晶粒時，將1個脈衝狀雷射 分割、延遲爲複數之分割雷射，將這些分割雷射予以重疊 之雷射加以照射之方法及裝置。此脈衝狀雷射係藉由疊合 複數的分割雷射來使脈衝的時間寬變廣。藉此，可使藉由 1個脈衝狀雷射的照射所致之材料的熔化後，結晶化進展 的時間變長，能使結晶的大小變大。藉由使結晶粒變大， Φ 可使薄膜電晶·體（以下，稱爲「TFT」。）的源極、汲極 間之晶粒邊界數減少，能使TFT的動作速度變快。 [專利文獻1]日本專利特開平8- 1 48423號公報 [專利文獻2]日本專利特開平6-5537號公報 【發明內容】 [發明所欲解決之課題] 在以藉由雷射照射使其結晶化之薄膜狀的矽製材料來 製作平面面板的情形，很重要的是，TFT的動作速度要快 (2) (2)1301293 ，更重要的特性是，要求面板整體的TFT特性的差異要小 。爲此，需要使結晶粒的大小變得均勻。 但是，專利文獻1、2中，欠缺對此點之考慮。當然 ，於製作平面面板時，也寄望產出要優異。 爲了使結晶粒的大小變得均勻，需要以最佳照射能量 密度，每一處所複數次，特別是1〇〜20次照射由脈衝（ 時間）狀雷射所形成之方形（空間）的線束來製作。此時 ，雷射的重複頻率爲數百Hz的關係，脈衝狀雷射係以數 ms間隔被照射。因此，藉由1脈衝的照射，矽熔化、結 晶化後，被冷卻至照射前的溫度，2脈衝以後之脈衝狀雷 射陸續被照射。 即藉由均化器來將脈衝（時間）狀激光雷射整形爲方 形（空間）之線束，以5〜10%之進給間距對薄膜之a-Si ，以最佳之能量密度，每一處照射1 〇〜2 0次，得以形成 均勻的幾乎與雷射的波長大小相等之粒徑的結晶。 可以獲得與雷射光的波長大小相等的結晶粒之原因， 被認爲係如日本專利特開平10-2561 52所記載般，以雷射 光的照射，半導體膜熔化、再結晶化後所形成的表面粗糙 ，成爲光散射（光分割）的起點。此表面粗糙可定性理解 爲，基本上係由於固液狀態之密度變化所引起，固化往橫 方向進展，於結晶粒成長時之固化的終點（粒界部）形成 凹凸之現象。並且，如對此半導體膜的粗糙表面再度照射 雷射光的話，於此凹凸部所被散射之散射光彼此相互千涉 ，於膜表面形成駐波。因此，於多數次照射中，於重複此 -6 - (3) (3)1301293 過程中，最終於半導體膜的表面可形成預定週期的凹凸（ J· Sipe，J.F. Young, J.S.Perston，and Η.M. van Driel, Phys. Rev.B27, 1141，1155，200 1 ( 1 983 ) ) 〇 然後，將脈衝狀雷射照射於材料時，藉由將雷射對材 料之入射角設爲從垂直線（法線）至Γ以上，可以使結晶 粒的大小變大（例如，專利公開2004-172424 )。如此所 製作的結晶粒，具有能均勻爲與照射的雷射之波長幾乎相 同大小的特徵。 然後，這些方法，爲了使照射次數成爲10〜20次， 需要使線束的進給速度（進給速度=線束寬/照射次數X雷 射的重複頻率）變慢，因此，存在有生產效率低落之技術 性課題。 此係於以往技術中，照射於材料之1個脈衝狀雷射的 脈衝波形，對於材料的熔化、固化，只作用爲單一之脈衝 雷射的關係，無法獲得照射次數之減少效果所引起。 另外，如使激光雷射的振盪器長時間運轉的話，脈衝 能量的平均値會有數％以上的變化。由此變化所造成，最 佳能量値的寬度窄時，照射之能量密度會有從最佳値偏離 之情形，因此，需要頻繁地量測、調整雷射’因此，也有 生產停止、生產性降低之課題。 即材料（P-Si膜）的結晶性與雷射光的能量密度大有 關係，能量密度太低、太高都無法良好地獲得。然後’如 以最佳能量密度來製造的話，可以獲得大約與雷射光2的 波長（又=3 0 8 n m )相等大小的結晶粒。 (4) 1301293 另外，以下將可以獲得前述之與雷射的波長幾乎相等 之大小的結晶粒徑之最佳照射能量密度的寬度稱爲「製程 餘裕」。即於不做成分割雷射的一般之以往方法中，照射 次數20次，製程餘裕値一般爲2〇mJ/cm2，於照射次數10 次以下時，製程餘裕爲零。 [解決課題之手段] ^ 申請專利範圍第1項之發明，係於對薄膜狀的矽製材 料（1 2、1 1 2 )複數次照射脈衝狀雷射來形成結晶粒時， 將脈衝狀雷射予以分割、延遲爲複數的分割雷射，來照射 材料（1 2、1 1 2 )之薄膜材料的結晶化方法，其特徵爲： 藉由複數的分割雷射之重疊所形成的1個脈衝狀雷射 的脈衝波形（2 )，係具有：最大的強度I之極大峰値（ 23)、及超過1/2之強度之至少1個極大峰値（22)、及 至少位於1個相鄰之極大峰値（22、23 )之間，且強度降 • 低爲1/2以下之極小點（21 )。 申請專利範圍第2項所記載之發明，係於對薄膜狀的 砂製材料（1 2、1 1 2 )複數次照射脈衝狀雷射來形成結晶 粒時’將脈衝狀雷射予以分割、延遲爲複數的分割雷射， 來照射材料（1 2、1 1 2 )之薄膜材料的結晶化方法，其特 徵爲： 藉由複數的分割雷射之重疊所形成的1個脈衝狀雷射 的脈衝波形（2 )，係至少具有2個預定之強度ία以上之 極大峰値（2 2、2 3 );且具有：強度從使材料（丨2、1 12 -8 - (5) 1301293 )產生熔化之極大峰値（22 )降低至極小點（2 i )，使材 料（1 2、1 1 2 )產生結晶化後，使材料（丨2、n 2 )產生熔 化之極大峰値（23 )。 申請專利範圍第3項所記載之發明，係如申請專利範 圍第1項之薄膜材料的結晶化方法，其中：極小點（2 i ) 的強度，係爲1/10以上。 申請專利範圍第4項之發明，係於對薄膜狀的矽製材 • 料（1 2、1 1 2 )複數次照射脈衝狀雷射來形成結晶粒時， 將脈衝狀雷射予以分割、延遲爲複數的分割雷射，來照射 材料（1 2、1 1 2 )之薄膜材料的結晶化方法，其特徵爲： 藉由複數的分割雷射之重疊所形成的1個脈衝狀雷射 的脈衝波形（2 )，係具有極大的強度I的極大峰値（23 ^ )，並且，具有1個1/1〇以上、1/2以下之強度的極小點 (21)，並且，於極小點（21)之前後，各存在一個超過 1/2之強度I、IA的極大峰値（22、23 )。 φ 申請專利範圍第5項之發明，係於對薄膜狀的矽製材 料（1 2、1 1 2 )複數次照射脈衝狀雷射來形成結晶粒時， 將脈衝狀雷射予以分割、延遲爲複數的分割雷射，來照射 材料（1 2、1 1 2 )之薄膜材料的結晶化方法，其特徵爲： 藉由複數的分割雷射之重疊所形成的1個脈衝狀雷射 的脈衝波形（3 )，係具有：由照射之開始點（30 )與第1 極小點（31 )間之第1極大峰値（41 )、及第1極小點（ 3 1 )與第2極小點（3 2 )間之第2極大峰値（42 )、及第 (N-1 )極小點（37 )與第N極小點（38 )間之第N極大 (6) 1301293 峰値（48 )、及第N極小點與照射之結束點（5(〇間之第 (N + 1 )極大峰値（4 9 )所構成，使材料（i 2、i i 2 )產生 結晶化之合計N個之極小點；及使材料（i 2、i i 2 )產生 熔化之（N + 1 )個之極大峰値， N爲3以上之整數， 第η極小點之強度，對於前面之第η極大峰値的強度 In’爲Ιη/10以上、Ιη/2以下之強度，η爲範圍之整 _ 數。 申請專利範圍第6項之發明，係如申請專利範圍第5 項之薄膜材料的結晶化方法，其中：第η極小點的強度， 對於之後的第η + 1極大峰値強度In，爲Ιη/1〇以上、Ιη/2 以下之強度。 申請專利範圍第7項之發明，係如申請專利範圍第1 、2、3、4、5或6項所記載之薄膜材料的結晶化方法，其 中：每一個前述材料（1 2、1 1 2 )的脈衝狀雷射之照射次 φ 數，爲1 〇次以下。 申請專利範圍第8項之發明，係如申請專利範圍第1 、2、3、4、5、6或7項所記載之薄膜材料的結晶化方法 ，其中：前述脈衝狀雷射並非直線偏光之雷射，入射光線 對材料（1 2、1 1 2 )之法線的入射角，爲Γ以上。 申請專利範圍第9項之發明，係於對薄膜狀的矽製材 料（1 2、1 1 2 )複數次照射來自雷射振盪器（i、1 〇丨）的 脈衝狀雷射（20、1 02 )來形成結晶粒時，將脈衝狀雷射 予以分割、延遲爲複數的分割雷射，來照射材料（1 2、 -10- (7) 1301293 1 1 2 )之薄膜材料的結晶化裝置，其特徵爲： 具有=使脈衝狀雷射延遲爲複數的分割雷射而予以分 割之分割手段；及 將複數的分割雷射之強度分布加以整形，使重疊照射 於材料（1 2、1 1 2 )之整形手段（丨丨〇、n丨）， 藉由複數的分割雷射之重疊所形成的i個脈衝狀雷射 的脈衝波形（2 )’係具有最大的強度I的極大峰値（23 )，並且，具有1個1/10以上、1/2以下之強度的極小點 (21)，並且，於極小點（21 )之前後，各存在一個超過 1/2之強度I、IA的極大峰値（22、23 )。 [發明效果] . 如依據獨立申請專利範圍第1、4、5及9，對薄膜狀 的矽製材料，複數次照射脈衝狀雷射（複數脈衝）來形成 結晶粒時，藉由照射於材料之分割雷射的重疊所形成的1 φ 個之脈衝狀雷射的脈衝波形，至少具有2個極大峰値，且 至少於2個極大峰値之間具有強度從最大強度I的極大峰 値降低製1/2以下之極小點。藉此，於最大的強度I之極 大峰値或其他的極大峰値中，使材料產生熔化，於2個極 大峰値之間的極小點中，可使材料結晶化。 藉由此種1個之脈衝狀雷射的分割雷射之照射，材料 之熔化、結晶化後，不會被冷卻至照射前的溫度，於結晶 化後，即刻再度進行熔化，可使脈衝狀雷射的照射次數減 少’能進行快速且均勻大小之結晶粒的結晶化。其結果可 -11 - (8) 1301293 使面板整體的TFT特性之差異變小。 然後，適當採用分割雷射的延遲時間，而使極小點的 強度成爲1/2以下的話，可以獲得能賦予適當的製程餘裕 之製程條件，可使材料良好地結晶化。 另外，如依據申請專利範圍第3、4、5、6及9般， 如使極小點的強度成爲1/ 1 0以上，可以避免脈衝狀雷射的 最佳結晶化能量的提高。 φ 如依據獨立申請專利範圍第2項，於薄膜狀的矽製材 料複數次照射脈衝狀雷射來形成結晶粒時，藉由照射於材 料之分割雷射的重疊所形成的1個脈衝狀雷射之脈衝波形 ，係至少具有2個預定強度IA以上的極大峰値，且具有 強度從使材料產生熔化之極大峰値降低至極小點，使材料 . 產生結晶化後，使材料產生熔化之極大峰値。 藉此，藉由1個脈衝狀雷射之分割雷射的照射，於材 料之熔化、結晶化後，不會被冷卻至照射前的溫度，結晶 φ 化後，即刻再度進行熔化。其結果爲可使脈衝狀雷射的照 射次數減少，能進行快速且均勻大小之結晶粒的結晶化。 脈衝狀雷射的照射次數，係如申請專利範圍第7項般 ，每一個材料，可以設爲1 〇次以下。 如依據申請專利範圍第8項，照射於材料之脈衝狀雷 射，並非直線偏光之雷射，入射光線對材料之法線的入射 角爲Γ以上，不用使用偏光板而使輸出降低，可使結晶粒 的大小變大。 -12 - (9) 1301293 【實施方式】 第1〜第7圖係表示關於本發明之薄膜材料的結晶化 裝置之1實施形態。第1圖中，符號1係表示雷射振盪器 ，具體而言，係射出不是直線偏光之紫外域雷射之激光雷 射（波長308nm)。從雷射振盪器1所被射出的脈衝狀雷 射20，係藉由透過率50 %之半透過反射鏡221而被分割爲 2個，以半透過反射鏡221所反射之第1分割雷射20a， ^ 係射入均化器224，另外，透過半透過反射鏡221之第2 分割雷射20b係於透過延遲電路223後，以全反射鏡222 而射入均化器224。藉由此半透過反射鏡221、全反射鏡 222及延遲電路223，構成將1個脈衝狀雷射相互延遲分 割爲複數的分割雷射之分割手段。 均化器224以後係與一般的以往方法相同，將第1、 第2分割雷射20a、20b形成爲線束，於薄膜狀的矽製材 料12重疊照射延遲時間：T超過0ns而在100ns以下之複 φ 數的脈衝之第1、第2分割雷射20a、20b而使結晶化。 於以適當的延遲時間T使作用爲第2 ( a )圖所示之2 個脈衝的第1、第2分割雷射20a、20b重疊而照射時，使 第1、第2分割雷射20a、20b重疊之脈衝波形2，係如第 2 ( b )圖所示般，於最大峰値2 3的強度爲I時，具有強 度從最大峰値23降低至1/10以上、1/2以下之極小點21 ，形成1個超過1/2之強度IA的極大峰値22。此極大峰 値23係藉由第1、第2分割雷射20a、20b的重疊，可使 材料12熔化者，於極小點21之後產生。另外，此脈衝波 -13- (10) 1301293 形2係於最大峰値2 3的強度爲I時’作爲可使材料12結 晶化者’形成1個1/10以上、1/2以下強度之極小點21， 於極小點2 1的前後，各形成丨個可使材料i 2熔化之超過 1/2強度的極大點22/ 23。極大點22、23及極小點21的 數目’係依據分割雷射20a、20b的分割數而改變。具有 複數個極大點22、23及極小點21，於以1/10以上' 1/2 以下之強度’可獲得結晶化之至少1個極小點21的前後 • ’各形成1個以超過1/2之強度使材料〗2熔化之極大點 2 2、2 3即可。 藉由將從此雷射振盪器1所射出的1個脈衝狀雷射2〇 如第1、第2分割雷射20a、20b般予以分割、延遲（延遲 時間：T = 0〜1 0 0 n s )爲複數後，使其重疊之脈衝波形2的 雷射照射於材料1 2之方法，嘗試使脈衝狀雷射20的照射 次數減少，來使結晶粒的大小變得均勻，單單使脈衝的時 間寬變廣’於照射次數之減少上，並無效果，得知只於對 鲁 桌2(a)圖所不之第1、第2分割雷射20a、20b給予預 定條件之情形，即使使照射次數減少，也可獲得均勻且大 的結晶。 即爲了使照射次數減少，使結晶粒的大小變得均勻， 於將1個脈衝狀雷射20分割、延遲爲複數後，使其重疊 來對材料12照射雷射時，雷射的脈衝波形2需要：可使 材料1 2熔化之強度I的最大峰値23、及可使材料1 2結晶 化之強度降低至1/2以下之至少1個極小點2 1、及可使材 料12熔化之超過1/2之強度的至少1個極大峰値22。即 -14- (11) 1301293 脈衝波形2係可有1個至少位於1個鄰接之極大峰値22、 2 3之間的極小點21，於極小點21的前後，各存在1個使 材料12熔化之極大峰値22、23。但是，最大峰値23係極 大峰値之一種，爲最大之強度I之極大峰値，所以，照射 於材料1 2之1個脈衝狀雷射的脈衝波形2，係至少具有2 個可使材料12熔化之強度IA以上的極大峰値22、23， 極小點21位於鄰接之2個極大峰値22、23間，只要具有 強度從使材料1 2熔化之極大峰値22降低至可使材料1 2 結晶化之極小點2 1，於使材料〗2產生結晶化後，使材料 12再度產生熔化之最大峰値23即可。 即將於把1個脈衝狀雷射20予以分割、延遲爲複數 後使其重疊之雷射照射於材料1 2時，雷射的脈衝波形2 於最大峰値2 3之強度爲I時，於獲得超過I / 2之強度的極 大峰値22後，產生強度降低至1/2以下之極小點21，之 後，能產生最大峰値23(極大峰値23)即可。將依序具 有此超過1/2之強度的極大峰値22、及強度降低至1/2以 下之極小點2 1、及最大峰値23 (超過1/2之強度的最大峰 値23 )之雷射的脈衝波形2稱爲第1脈衝波形。 然後’於對薄膜狀的矽材料12複數次照射（複數脈 衝照射）脈衝狀雷射20來形成結晶粒之方法中，將具有 第1脈衝波形之雷射照射於材料1 2來使薄膜之材料1 2結 晶化的話，即使使從雷射振盪器1所射出而照射之脈衝狀 雷射20的數目減少，也可以獲得均勻之結晶。其結果爲 ，結晶粒的大小之差異變小，面板整體的TFT特性之差異 -15- (12) 1301293 變小。 說明結晶之成長過程。結晶之成長可以考慮爲，以第 1次之照射所產生的結晶粒，會藉由第2次以後之照射而 結合變大。於此結晶之成長上，需要再度實施雷射之照射 ，使材料1 2從被冷卻而固化，即結晶化之狀態升溫至熔 化溫度附近，來使其熔化、再結晶化。雷射之能量密度如 爲最佳的話，結晶粒會成長，可以獲得幾乎與雷射之波長 g 相等之大小的結晶粒。 對此材料1 2給予熔化、再結晶化之最佳的能量密度 ，於單獨之脈衝狀雷射20具有之情形外，於脈衝波形2 具有第1脈衝波形之情形，也可以獲得。 於脈衝波形2具有第1脈衝波形之情形，矽製之材料 12在第1峰値22附近熔化，材料12於強度降低至1/2以 下之極小點21附近開始結晶化，之後，藉由第2峰値23 射入’材料12再度熔化、結晶化。如此，藉由分割雷射 φ 於材料1 2之結晶化開始後給予熱量，認爲可以獲得具有 不做成分割雷射之以往的只1個之峰値所構成的脈衝波形 的脈衝狀雷射，予以隔以充分時間而加以2次照射之情形 相同的效果。 爲了確認此效果，將由半値寬=25ns所形成之1個脈 衝狀雷射20如第2 ( a )圖所示般予以分割，獲得第i、 第2分割雷射20a、20b，將其整形爲線束，成爲延遲時間 :T = 0~100ns之2個脈衝之第1、第2分割雷射20a、20b 予以重疊照射於薄膜狀的矽製材料1 2，使其結晶化。如第 -16- (13) 1301293 2 ( a )圖所示般，延遲時間τ之第1、第2分割雷射 、20b的各半値寬爲W。 使第1、第2分割雷射20a、20b重疊之脈衝波形 係如第2 ( b )圖所示般，於最大峰値23之強度爲I 具有強度從最大峰値23降低製1/10以上、1/2以下之 點21，超過1/2之強度IA的極大峰値22爲1個。此 峰値23係藉由第1、第2分割雷射20a、20b之重疊 | 極小點2 1之後產生。 然後，量測由第1、第2分割雷射20a、20b (脈 形2 )所形成的脈衝狀雷射20之對材料1 2的照射次I 、20次中之製程餘裕，與將2個分割雷射20a、20b 合倂計算時之最佳結晶化能量密度，獲得第4、第5 結果。極小點21係使延遲時間T變長的話，會變小 T = 30ns時，成爲1/2，於T = 50ns時，成爲1/10。因 如之後敘述般，以將半値寬W = 25ns設定爲比延遲時 φ 小，並且設定延遲時間T = 30〜50ns爲佳。另外，脈衝 射20之脈衝能量：670mJ，重複頻率：300Hz，第4 照射次數（1 0次、20次）係脈衝狀雷射20之照射次| 依據第4圖、第5圖’於一般的以往方法中（延 間T = 0 n s )，在照射次數1 〇次中，無製程條件（製 裕爲零）。如使分割雷射2 0 a、2 0 b之延遲時間Τ逐 長，從延遲時間T = 20ns以上開始產生正的製程餘裕 由設爲延遲時間T = 30ns以上（極小點的強度1/2以下 可以獲得適當的製程條件（適當的製程餘裕）。因此 20a 2， 時， 極小 最大 ，於 衝波 ^ 10 予以 圖之 ，於 此， 間T 狀雷 圖之 致。 遲時 程餘 漸變 ，藉 )， ，將 -17- (14) 1301293 極小點21的強度，即極小値設爲1/2以下。但是，由第5 圖可以明白，設延遲時間T = 5〇ns以上（極小點的強度 1/ 1 0以下），則最佳結晶化能量開始急遽上升，設爲極小 點21的強度1/10以上。 如第4圖所示之極小値1/2以下之製程餘裕（照射次 數10次，約20mJ/cm2，照射次數20次，約40mJ/cm2 ) ’與前述之不做成分割雷射之一般的以往方法的製程餘裕 p (照射次數10次爲零，照射次數20次，約20mJ/cm2)比 較，有所提升。 以第4圖之照射次數1 0次來結晶化之矽薄膜製的材 料12，藉由SEM (掃描型電子顯微鏡）來觀察，測量1 視野內之結晶粒的大小，以結晶粒的90%以上爲雷射波長 之大小308nm±30nm時爲製程條件，來觀察可以獲得製程 條件之照射能量密度的容許寬（製程餘裕）。其結果爲， 照射次數：10次時之製程餘裕（約20mJ/cm2 )，與不做 φ 成分割雷射之一般的以往方法的照射次數20次來結晶化 之材料，幾乎相同，可以良好地獲得幾乎同樣大小之結晶 粒。 如此，藉由將極小點的強度値設爲1/2以下，知道具 有能降低照射次數之效果，接著，設爲整數N - 3，製作 極大峰値爲N+1個、滿足1/2〜1/10之極小點爲N個，藉 由試驗來確認其效果。 藉由此N + 1個之分割雷射的重疊所形成之1個脈衝狀 雷射的脈衝波形3，係如第3圖所不般’如照射之開始點 -18- (15) 1301293 3〇與第1極小點3 i間之第1極大峰値4 1、第1極小點3 1 與第2極小點32間之第2極大峰値42、第（N-1 )極小 點37與第N極小點38間之第N極大峰値48、第N極小 點與照射之結束點50間之第（N + 1 )極大峰値49般，具 有使材料1 1 2產生結晶化之合計N個的極小點與使材料 1 1 2產生熔化之（n+ 1 )個之極大峰値。然後，結晶化方 法設爲：第η極小點的強度，係對之前的第η極大峰値之 | 強度In，爲Ιη/ΐο以上、Ιη/2以下之強度的結晶化方法， 或者第η極小點的強度，係對之後的第η+ 1極大峰値的強 度Ιη+1，爲In/ίο以上、Ιη/2以下之範圍的強度之結晶化 方法，對於材料1 1 2，藉由1個脈衝狀雷射，至少依序使 其熔化、結晶化、熔化、結晶化，可以使照射次數減少。 但是，η爲1〜Ν之範圍的整數。 爲此之試驗裝置，係使用第7(A) 、( Β ) 、( C ) 圖所示之複數（圖上爲3個）之半透過反射鏡103、104、 φ 1〇5及1個全反射鏡106，將從振盪器101所射出的1個 脈衝狀雷射102予以分割，以延遲時間T = 30ns (極小點 的強度値1/2 )來形成時間接近之複數（圖上爲4個）之 相鄰的分割雷射120a、120b、120c、120d，將全部的分割 雷射120a、120b、120c、120d射入共通的均化器110。因 此，將延遲電路 107、108、109配置於各半透鏡反射鏡 103〜105及全反射鏡106之間。藉此，製作第3圖所示之 極大峰値爲N+1個、極小點爲N個之脈衝波形。藉由此 半透過反射鏡103〜105、全反射鏡106及延遲電路107〜 -19- (16) 1301293 1 09，構成使脈衝狀雷射1 02相互延遲分割爲複數的分割 雷射120a、120b、120c、120d之分割手段。第3圖所示 之脈衝波形3的鄰接之2個極大峰値之強度In、ιη+ι，只 要超過其間之極小點的強度In/2之強度，可使材料1 1 2熔 化即可，於複數的強度In之極大峰値之中存在有最大強 度之極大峰値。 實際上，改變半透過反射鏡103〜105及延遲電路1〇7 | 〜1 09之個數，將極大峰値之數目爲1〜7之脈衝波形照射 於薄膜狀的矽製之材料1 2，測量脈衝狀雷射之照射次數5 、1 0、1 5、20次之極大峰値數與製程餘裕的關係。將其結 果表示於第6圖。依據第6圖，極大峰値數（及極小點數 )如變多的話，則製程餘裕擴大，即使照射次數爲5次， 如做成使極大峰値數成爲3以上之分割雷射的話，知道可 以獲得良好的製程條件。此照射次數係脈衝狀雷射1 02的 ' 數目。每一個材料12的脈衝狀雷射之照射次數，可在1 〇 φ 次以下即很充分。 [實施例] 參照第7圖來具體說明1實施例。 將藉由 XeCl激光雷射振盪器101，以脈衝能量： 670mJ、重複頻率：300Hz所射出之脈衝波形的脈衝狀雷 射102藉由反射率爲Rl、R2、R3、R4之4個反射鏡1〇3 、104、105、106予以分割、方向轉換，藉由設置於前述 反射鏡103、104、105、106間之延遲電路107、108、1〇9 -20- (17) 1301293 使鄰接之分割雷射120a〜120d彼此時間延遲，陸續射入長 軸均化器1 1 0，以周知手段來整形爲長軸：200mm長之矩 形波的空間強度分布，照射於材料1 2之玻璃上的膜厚 50nm之a-Si膜II2。反射鏡103之反射率爲R1、反射鏡 104的反射率爲R2、反射鏡105之反射率爲R3、全反射 鏡106的反射率爲R4=100%。 此時，將前述反射率設爲：Rl=25%、R2 = 33.3%、 R3 = 50%、R4 = l〇〇%，以獲得脈衝波形爲極大峰値強度可以 週期30ns產生4個之第3圖的波形3。 激光雷射係發出對a-Si膜1 12之吸收力高之紫外區域 的非直線偏光之光線’可以獲得局輸出之脈衝光（波長 3 0 8 n m ) 〇 另外，各延遲電路107、108、109係具備使產生光路 差所需之複數的全反射鏡，藉由這些全反射鏡，來對分割 雷射120a〜120d設置時間差者，爲了進行個別延遲時間： 3 0ns之延遲，設置9m之光路。 於與脈衝狀雷射1 〇 2的長軸正交之方向，設置如第7 (C )圖所示之短軸均化器1 1 1，整形爲短軸寬爲0.4mm 之矩形波的空間強度分布，利用入射光線對材料1 1 2之法 線的入射角α = 5 °來照射於材料1 12。另外，玻璃上之a-Si膜112係載於未圖示出之載台，以速度：V( mm/s )在 短軸方向掃描。然後，依據V = 0.4 · 300/Z，設爲V = 24、 12、8、6mm/s，來使得a - S i膜1 1 2之一處的照射次數： Z = 5、10、15、20次。均化器110、111係構成將複數的 -21 - (18) l3〇1293 分割雷射之強度分布予以整形，使重疊照射於材料1 1 2之 整形手段。 脈衝狀雷射1 02的照射後之結晶化的矽薄膜製之材料 112，藉由SEM (掃描型電子顯微鏡）來觀察，測量1視 野內之結晶粒的大小，以結晶粒的90%以上爲雷射波長之 大小308nm±30nm時爲製程條件，來測量可以獲得製程條 件之照射能量密度的容許寬（製程餘裕）。其結果爲，照 ί 射次數：5、10、15、20次時之製程餘裕，係如第6圖表 示爲極大峰値數4般，約20、40、60、lOOmJ/cm2，即使 於以往製程爲不可行之照射次數：5次以下，也可以獲得 能成爲製程之條件。 藉由將脈衝狀雷射102 (分割雷射120 a〜120d)對材 料1 1 2之入射角α設爲對於垂直線（法線）爲Γ以上，知 、 道可使結晶粒比雷射1 02的波長更大。因此，將入射光線 對材料1 1 2的法線之入射角α設爲Γ以上。另外，雷射之 I 重複頻率爲數百Hz的關係，脈衝狀雷射102係以數ms間 隔被照射。因此，藉由複數脈衝之脈衝狀雷射的照射，矽 (材料1 1 2 )重複熔化、冷卻結晶化，成爲照射之雷射的 波長幾乎相同或其以上之大小而均勻，所以，可使結晶粒 的大小之差異變小。將1個之脈衝狀雷射1 02當成複數的 分割雷射120a〜120d來照射的話，認爲可以藉由1個之脈 衝狀雷射1 02，來獲得與照射因應分割數之複數脈衝的脈 衝狀雷射同等的效果。 另外，如日本專利特開2004- 172424所記載般，照射 -22- (19) 1301293 材料1 1 2之脈衝狀雷射i 〇 2如非直線偏光的雷射，可利用 偏光板’不使輸出降低而使結晶粒成長，另外，入射光線 對材料之法線的入射角如爲Γ以上的話，藉由脈衝狀雷射 1 02的重複照射，結晶粒可與雷射丨〇2的波長相等或比雷 射的波長還大。激光雷射爲不具有特定的偏光之隨機偏光 。將入射光線對材料之法線的入射角設爲Γ以上，可以獲 得大的結晶粒。 【圖式簡單說明】 第1圖係表示關於本發明之1實施形態之薄膜材料的 結晶化裝置之槪略圖。 第2圖係同樣地表示以橫軸表時間、縱軸表強度之分 . 割雷射，第2 ( a )圖係個別表示分割雷射之線圖，第2 ( b)圖係表示分割雷射之重疊狀態之線圖。 第3圖係同樣地表示以橫軸表時間、縱軸表強度之分 φ 割雷射的重疊狀態之線圖。 第4圖係同樣地表示延遲時間-製程餘裕特性之線圖 〇 第5圖係同樣地表示延遲時間-最佳結晶化能量特性 之線圖。 第6圖係同樣地表示極大峰値數-製程餘裕特性之線 圖。 第7圖係同樣地表示1實施例所使用之薄膜材料的結 晶化裝置，第7 ( A )圖係表示雷射振盪器及分割手段之 •23- (20) 1301293 槪略圖’弟7(B)圖係表示省略 化器製材料之槪略圖，第7(C) 製材料之槪略圖。 【主要元件之符號說明】 1、 1 0 1 :雷射振盪器， 2、 3 :脈衝波形， 1 2、1 1 2 :薄膜狀的矽製材料， 20、 102 :脈衝狀雷射， 20a 、 20b 、 120a 、 120b 、 120c 、 21、 31、 32、 37、 38 ：極小點， 22、 23、41、42、48、49: @ -) 3 0 :開始點， 5 0 :結束點， 103、104、105 :半透過反射鏡 106 :全反射鏡（分割手段）， 107、108、109 :延遲電路（分 1 1 0、1 1 1 :均化器（整形手段） 221 :半透過反射鏡（分割手段 222 :全反射鏡（分割手段）， 223 :延遲電路（分割手段）。 軸均化器，由長軸均 係表示從短軸均化器 1 2 0 d :分割雷射， C峰値， (分割手段）， 割手段）， )， -24 -1301293 (1) _ IX IX. OBJECTS OF THE INVENTION [Technical Field] The present invention relates to a method for crystallizing a film material and an apparatus therefor, and more particularly to irradiating a pulsed laser onto a film-like tantalum material, A method of crystallizing a film material used for the production of a crystallized crucible for a thin film transistor used for a flat panel, and a device thereof. B. [Prior Art] Patent Literatures 1 and 2 disclose that when a film-shaped tantalum material is irradiated with a plurality of pulsed lasers to form crystal grains, one pulsed laser is divided and delayed into a plurality of divided rays. A method and apparatus for irradiating these divided lasers with overlapping lasers. This pulsed laser broadens the time width of the pulse by superimposing a plurality of split lasers. Thereby, the melting of the material by irradiation of one pulsed laser can increase the time for crystallization to progress, and the size of the crystal can be increased. By increasing the crystal grains, Φ can reduce the number of grain boundaries between the source and the drain of the thin film transistor (hereinafter referred to as "TFT"), and the operating speed of the TFT can be increased. [Patent Document 1] Japanese Laid-Open Patent Publication No. Hei. No. Hei. No. Hei. No. Hei. No. Hei. In the case of a crystallized film-like tantalum material to produce a flat panel, it is important that the TFT be operated at a faster speed (2) (2) 1301293. A more important characteristic is that the difference in TFT characteristics of the panel is required to be different. small. For this reason, it is necessary to make the size of the crystal grains uniform. However, in Patent Documents 1 and 2, the consideration of this point is lacking. Of course, when making flat panels, I also hope that the output will be excellent. In order to make the size of the crystal grains uniform, it is necessary to irradiate the square (space) wire bundle formed by the pulse (time) laser at a plurality of times, in particular, 1 to 20 times, at an optimum irradiation energy density. Production. At this time, the repetition frequency of the laser is in the relationship of several hundred Hz, and the pulsed laser is irradiated at intervals of several ms. Therefore, after one pulse of irradiation, the crucible is melted and crystallized, and then cooled to a temperature before irradiation, and pulsed lasers after two pulses are successively irradiated. That is, by means of a homogenizer, the pulsed (time) laser beam is shaped into a square (space) wire bundle, with a feed pitch of 5 to 10% for the a-Si of the film, with an optimum energy density, each Irradiation of 1 〇~2 0 times produces a uniform crystal of a particle size almost equal to the wavelength of the laser. The reason why the crystal grain of the same size as the wavelength of the laser light can be obtained is considered to be a surface formed by melting and recrystallizing the semiconductor film by irradiation of laser light as described in Japanese Patent Laid-Open No. Hei 10-2561 52. Roughness becomes the starting point of light scattering (light segmentation). This surface roughness can be qualitatively understood to be basically caused by a change in the density of the solid-liquid state, and the solidification progresses in the transverse direction, and the end point (grain boundary portion) of the solidification at the time of growth of the crystal grains forms a concavity and convexity. Further, when the rough surface of the semiconductor film is irradiated with the laser light again, the scattered light scattered by the uneven portion interferes with each other to form a standing wave on the surface of the film. Therefore, in the majority of the irradiation, in the process of repeating this -6 - (3) (3) 1301293, a predetermined period of irregularities can be formed on the surface of the semiconductor film (J. Sipe, JF Young, JSPerston, and Η. M. van Driel, Phys. Rev. B27, 1141, 1155, 200 1 ( 1 983 ) ) 〇 Then, when the pulsed laser is irradiated onto the material, the incident angle of the laser is set to be from the vertical line. (Normal) to above, the size of the crystal grains can be made large (for example, Patent Publication No. 2004-172424). The crystal grains thus produced have a feature that they are uniformly uniform in size to the wavelength of the irradiated laser. Then, in order to make the number of times of irradiation 10 to 20 times, it is necessary to slow down the feed rate of the wire harness (feed rate = wire bundle width / number of irradiations X repetition frequency), and therefore, there is a low production efficiency. Technical issues. In the prior art, the pulse waveform of one pulsed laser irradiated on the material causes only a single pulsed laser to melt and solidify the material, and the effect of reducing the number of irradiations cannot be obtained. Further, if the laser laser oscillator is operated for a long period of time, the average enthalpy of the pulse energy may vary by several % or more. As a result of this change, when the width of the optimal energy 窄 is narrow, the energy density of the irradiation will deviate from the optimum enthalpy. Therefore, it is necessary to frequently measure and adjust the laser. Therefore, production is stopped and productivity is lowered. The subject. That is, the crystallinity of the material (P-Si film) is greatly related to the energy density of the laser light, and the energy density is too low and too high to be well obtained. Then, if it is manufactured at an optimum energy density, crystal grains having a size approximately equal to the wavelength of the laser light 2 (again, 3 0 8 n m ) can be obtained. (4) 1301293 In addition, the width of the optimum irradiation energy density of the crystal grain size which is almost equal to the wavelength of the laser described above can be obtained as "process margin". That is, in the conventional conventional method in which the laser beam is not divided, the number of irradiations is 20, and the process margin is generally 2 〇 mJ/cm 2 , and when the number of times of irradiation is 10 or less, the process margin is zero. [Means for Solving the Problem] ^ The invention of claim 1 is based on the case where a film-like tantalum material (1 2, 1 1 2) is irradiated with a pulsed laser to form crystal grains. A method of crystallization of a thin film material which is divided into a plurality of divided lasers and irradiated with a plurality of divided lasers to irradiate a material (1 2, 1 1 2 ), characterized by: a pulse formed by overlapping of a plurality of divided lasers The pulse waveform (2) of the laser has: a maximum intensity I (23) of maximum intensity I, and at least one maximum peak (22) of intensity exceeding 1/2, and at least one adjacent Between the extreme peaks (22, 23) and the intensity drop • is a minimum of 1/2 or less (21). In the invention described in the second aspect of the patent application, when a film-shaped sand material (1 2, 1 1 2 ) is irradiated with a pulsed laser to form crystal grains, the pulsed laser is divided and delayed. A method of crystallization of a film material for irradiating a material (1 2, 1 1 2 ) for a plurality of divided lasers, characterized by: a pulsed laser pulse formed by overlapping of a plurality of divided lasers The waveform (2) is a maximum peak 値 (2 2, 2 3 ) having at least two predetermined intensities ία or more; and has: the intensity is melted from the material (丨2, 1 12 -8 - (5) 1301293) The maximum peak 値(22) is reduced to a minimum point (2 i ), and after the material (1 2, 1 1 2 ) is crystallized, the material (丨2, n 2 ) is subjected to a melting peak (23). The invention described in claim 3 is a method for crystallizing a film material according to the first aspect of the patent application, wherein the intensity of the minimum point (2 i ) is 1/10 or more. The invention of claim 4 is based on the fact that when a film-like enamel material (1 2, 1 1 2) is irradiated with a pulsed laser to form crystal grains, the pulsed laser is divided and delayed. A method of crystallization of a film material of a plurality of divided lasers to illuminate a material (1 2, 1 1 2 ), characterized by: a pulse waveform of a pulsed laser formed by overlapping of a plurality of divided lasers (2) is a very large peak (23^) having a maximum intensity I, and has a minimum point (21) of 1/1 〇 or more and 1/2 or less intensity, and is at a minimum point (21). Before and after, there is a maximum peak 値 (22, 23) of more than 1/2 of the intensity I, IA. φ The invention of claim 5 is based on the fact that when a film-like tantalum material (1 2, 1 1 2 ) is irradiated with a pulsed laser to form crystal grains, the pulsed laser is divided and delayed. A method of crystallization of a film material of a plurality of divided lasers to illuminate a material (1 2, 1 1 2 ), characterized by: a pulse waveform of a pulsed laser formed by overlapping of a plurality of divided lasers (3) has a first maximum peak 値(41) between the start point (30) of the irradiation and the first minimum point (31), and a first minimum point (3 1 ) and a second minimum point (3) 2) The 2nd maximum peak 値(42), and the Nth largest point (6) 1301293 値(48) between the (N-1)th smallest point (37) and the Nth minimum point (38), and The N-minimum point and the end point of the irradiation (5 (N + 1) maximum peak 値 (4 9 ) are formed, so that the material (i 2, ii 2 ) produces a total of N minimum points of crystallization; And causing the material (i 2, ii 2 ) to produce a (N + 1 ) maximum peak 熔化, N is an integer of 3 or more, the intensity of the η-th pole, and the intensity of the η-max peak In in front The intensity of Ιη/10 or more and Ιη/2 or less, η is the integer _ number of the range. The invention of claim 6 is a crystallization method of the film material according to the fifth item of the patent application, wherein: η is extremely small The intensity of the point, for the subsequent η + 1 maximum peak intensity In, is Ι η / 1 〇 or more, Ι η / 2 or less. The invention of claim 7 is as claimed in claims 1 and 2. The method for crystallizing a film material according to item 3, 4, 5 or 6, wherein: the number of times of irradiation of the pulsed laser of each of the aforementioned materials (1 2, 1 1 2) is 1 or less. The invention of claim 8 is the method for crystallizing a film material as described in claim 1, 2, 3, 4, 5, 6 or 7 wherein the pulsed laser is not a linear polarized light. The incident angle of incident light to the normal of the material (1 2, 1 1 2 ) is above Γ. The invention of claim 9 is based on a film-like tantalum material (1 2, 1 1 2 a plurality of pulsed lasers (20, 102) from a laser oscillator (i, 1 〇丨) In the case of crystal grains, a crystallization device for dividing a pulsed laser into a plurality of divided lasers to irradiate a material of the material (1 2, -10-(7) 1301293 1 1 2 ) is characterized in that : a dividing means for dividing a pulsed laser delay into a plurality of divided lasers; and shaping the intensity distribution of the plurality of divided lasers so that the overlapping illumination is applied to the material (1 2, 1 1 2 ) Means (丨丨〇, n丨), a pulse waveform (2) of i pulsed lasers formed by overlapping of a plurality of divided lasers has a maximum peak 値 (23) having a maximum intensity I, and It has a minimum point (21) of 1/10 or more and 1/2 or less intensity, and after the minimum point (21), there is a maximum peak 超过 of intensity I and IA exceeding 1/2 ( 22, 23). [Effect of the Invention] According to the patent application scopes 1, 4, 5 and 9, the film-like tantalum material is irradiated with a plurality of times by irradiating a pulsed laser (complex pulse) to form crystal grains. The pulse waveform of 1 φ pulsed laser formed by the overlap of the divided lasers has at least two maximum peaks, and at least the maximum peak 至少 is reduced from the maximum intensity I between at least two maximum peaks The minimum point is 1/2 or less. Thereby, the material is melted in the extremely large peak of the maximum intensity I or other extreme peaks, and the material can be crystallized in a very small point between the two extremely large peaks. By the irradiation of the split laser of such a pulsed laser, after the material is melted and crystallized, it is not cooled to the temperature before the irradiation, and immediately after the crystallization, the melting is performed again, and the pulse can be formed. The number of irradiations of the laser is reduced to enable crystallization of crystal grains of a rapid and uniform size. As a result, -11 - (8) 1301293 makes the difference in TFT characteristics of the entire panel small. Then, if the delay time of the split laser is appropriately used and the intensity of the minimum point is 1/2 or less, a process condition capable of imparting an appropriate process margin can be obtained, and the material can be crystallized favorably. Further, as in the case of the third, fourth, fifth, sixth and ninth aspects of the patent application, if the intensity of the minimum point is 1/10 or more, the improvement of the optimum crystallization energy of the pulsed laser can be avoided. φ According to item 2 of the independent patent application, when a film-like tantalum material is irradiated with a pulsed laser light to form crystal grains, a pulse-like mine formed by the overlapping of the lasers irradiated by the material is formed. The pulse waveform of the shot has at least two maximum peaks of a predetermined intensity IA or more, and the intensity is reduced from a maximum peak of melting of the material to a minimum point, so that the material is crystallized and the material is greatly melted. Peak. Thereby, after the material is melted and crystallized by the irradiation of the laser beam by one pulsed laser, it is not cooled to the temperature before the irradiation, and the crystal is φ, and then immediately melted. As a result, the number of irradiations of the pulsed laser can be reduced, and crystallization of crystal grains of a rapid and uniform size can be performed. The number of irradiations of the pulsed laser is as in the seventh item of the patent application, and each material can be set to be less than 1 time. According to the eighth paragraph of the patent application scope, the pulsed laser irradiated on the material is not a linearly polarized laser, and the incident angle of the incident light to the normal of the material is Γ or more, and the output is lowered without using a polarizing plate. The size of the crystal grains becomes large. -12 - (9) 1301293 [Embodiment] Figs. 1 to 7 show an embodiment of a crystallization apparatus for a film material of the present invention. In Fig. 1, reference numeral 1 denotes a laser oscillator, specifically, a laser laser (wavelength 308 nm) which emits ultraviolet laser light which is not linearly polarized. The pulsed laser beam 20 emitted from the laser oscillator 1 is divided into two by a half-transmission mirror 221 having a transmittance of 50%, and the first divided laser beam reflected by the semi-transmissive mirror 221 is divided into two. 20a, ^ is incident on the homogenizer 224, and the second divided laser 20b transmitted through the semi-transmissive mirror 221 is transmitted through the delay circuit 223, and then incident on the homogenizer 224 by the total reflection mirror 222. The semi-transmissive mirror 221, the total reflection mirror 222, and the delay circuit 223 constitute a division means for dividing the pulse lasers into a plurality of divided laser beams. The homogenizer 224 is formed in the same manner as the conventional conventional method, and the first and second divided laser beams 20a and 20b are formed into a wire harness, and the film-shaped twisted material 12 is overlapped with the irradiation delay time: T exceeds 0 ns and is less than 100 ns. The first and second divided lasers 20a and 20b of the pulse of the complex number of φ are crystallized. When the first and second divided lasers 20a and 20b acting as the two pulses shown in the second (a) diagram are superimposed and irradiated with an appropriate delay time T, the first and second divided lasers 20a are placed. The pulse waveform 2 superimposed on 20b is as shown in the second (b) diagram. When the intensity of the maximum peak 値 2 3 is 1, the intensity decreases from the maximum peak 値 23 to 1/10 or more and 1/2 or less. The minimum point 21 forms a maximum peak 値22 of more than 1/2 of the intensity IA. This maximum peak 23 is caused by the overlap of the first and second divided lasers 20a and 20b, so that the material 12 can be melted and generated after the minimum point 21. In addition, the pulse wave-13-(10) 1301293-shaped 2 is formed when the intensity of the maximum peak 値2 3 is I, and the crystallization of the material 12 is formed to have one strength of 1/10 or more and 1/2 or less. The minimum point 21, before and after the minimum point 2 1 , forms a maximum point 22/23 which allows the material i 2 to melt more than 1/2 intensity. The number of the maximum points 22, 23 and the minimum points 21 is changed in accordance with the number of divisions of the divided laser beams 20a, 20b. There are a plurality of maximum points 22, 23 and a minimum point 21, and at least one minimum point 21 of crystallization can be obtained at a strength of 1/10 or more '1/2 or less. The strength of 2 makes the material 〗 2 melt the maximum point 2 2, 2 3 can be. The pulsed laser beam 2 emitted from the laser oscillator 1 is divided and delayed as in the first and second divided laser beams 20a and 20b (delay time: T = 0 to 1 0 0 ns ) After the plural, the laser beam of the pulse waveform 2 which is superimposed is irradiated onto the material 12, and the number of times of irradiation of the pulsed laser 20 is reduced to make the size of the crystal grain uniform, and the pulse time is simply made wide. In the case of the reduction of the number of times of irradiation, there is no effect, and it is known that only the first and second divided lasers 20a and 20b of the Lu table 2 (a) are given predetermined conditions, even if the number of times of irradiation is made By reducing, uniform and large crystals can also be obtained. In other words, in order to reduce the number of irradiations, the size of the crystal grains is made uniform, and when one pulsed laser beam 20 is divided and delayed into a plurality of numbers, and the material 12 is irradiated with a laser beam, the laser pulse waveform 2 is irradiated. It is required that the maximum peak 値 23 of the strength I at which the material 12 is melted, and at least one minimum point 2 1 which can reduce the strength of crystallization of the material 12 to less than 1/2, and the melting of the material 12 can be exceeded. At least 1 maximum peak 値 22 of 1/2 intensity. That is, the -14-(11) 1301293 pulse waveform 2 system may have a minimum point 21 at least between one adjacent maximum peaks 22, 2 3, and one material 12 is present before and after the minimum point 21 The maximum peak of melting is 22, 23. However, the maximum peak 値23 is one of the maximum peaks and is the maximum peak of the maximum intensity I. Therefore, the pulse waveform 2 of one pulsed laser irradiated to the material 12 has at least two materials. 12 melting intensity IA above the maximum peaks 22, 23, the minimum point 21 is located between the two adjacent extreme peaks 22, 23, as long as the intensity is reduced from the maximum peak 値 22 melting the material 12 to the material 1 2 The minimum point of crystallization is 2 1, after the material 〗 2 is crystallized, the material 12 is again allowed to produce the maximum peak 熔化23 of melting. When a pulsed laser 20 is divided and delayed to a complex number and the laser beam is superimposed on the material 12, the pulse waveform 2 of the laser is obtained when the intensity of the maximum peak 値 2 3 is When the maximum peak 値 22 exceeding the intensity of I / 2 is generated, the intensity is reduced to a minimum point 21 of 1/2 or less, and thereafter, a maximum peak 値 23 (maximum peak 値 23) can be generated. The maximum peak 値 22 having an intensity exceeding 1/2, and the minimum point 2 1 whose intensity is reduced to 1/2 or less, and the maximum peak 値 23 (the maximum peak 超过 23 exceeding the intensity of 1/2) will be The pulse waveform 2 of the laser is referred to as a first pulse waveform. Then, in the method of forming a crystal grain by a plurality of times of irradiation (multiple pulse irradiation) of the pulsed laser 20 on the film-like ruthenium material 12, a laser having a first pulse waveform is irradiated onto the material 12 to make a material of the film. In the case of crystallization, even if the number of the pulsed lasers 20 irradiated from the laser oscillator 1 is reduced, uniform crystals can be obtained. As a result, the difference in the size of the crystal grains becomes small, and the difference in the TFT characteristics of the entire panel -15-(12) 1301293 becomes small. Explain the growth process of crystallization. It is considered that the growth of crystals is such that the crystal grains generated by the first irradiation are combined by the second and subsequent irradiations. In the growth of the crystal, it is necessary to irradiate the laser again, and the material 12 is heated and solidified, i.e., crystallized, to a temperature near the melting temperature to be melted and recrystallized. If the energy density of the laser is optimal, the crystal grains will grow, and crystal grains having a size almost equal to the wavelength g of the laser can be obtained. The optimum energy density for melting and recrystallization of the material 12 is obtained in the case where the pulse waveform 2 has the first pulse waveform, in addition to the case of the single pulsed laser 20. In the case where the pulse waveform 2 has the first pulse waveform, the tantalum material 12 is melted in the vicinity of the first peak 22, and the material 12 starts to crystallize near the minimum point 21 whose intensity is reduced to 1/2 or less. 2 peak 値 23 injection 'material 12 is melted again and crystallized. In this way, by dividing the laser φ and applying heat to the start of crystallization of the material 12, it is considered that a pulsed laser having a pulse waveform composed of only one peak which is not a conventional laser beam can be obtained. The same effect is obtained in the case where the irradiation is performed twice in a sufficient time. In order to confirm this effect, one pulsed laser beam 20 formed by a half width = 25 ns is divided as shown in the second (a) diagram, and the i-th and second-divided laser beams 20a and 20b are obtained and shaped into The wire harness is a delay time: the first and second divided laser beams 20a and 20b of two pulses of T = 0 to 100 ns are superimposed and irradiated on the film-shaped tantalum material 1 2 to be crystallized. As shown in the figure -16-(13) 1301293 2 (a), the widths of the first and second divided lasers of the delay time τ and the half widths of 20b are W. The pulse waveform in which the first and second divided laser beams 20a and 20b are superimposed is as shown in the second (b) diagram, and the intensity at the maximum peak 値 23 is I, and the intensity is reduced from the maximum peak 値 23 by 1/10 or more. At a point 21 of 1/2 or less, the maximum peak 値 22 of the intensity IA exceeding 1/2 is one. This peak 23 is generated by the overlap of the first and second divided lasers 20a and 20b and the minimum point 2 1 . Then, the irradiation time I of the pair of pulsed lasers 20 formed by the first and second divided lasers 20a and 20b (pulse 2) is measured, and the process margin is 20 times, and 2 The optimal crystallization energy density at the time of splitting the laser 20a and 20b is calculated, and the fourth and fifth results are obtained. The minimum point 21 is such that the delay time T becomes longer, and becomes smaller when T = 30 ns, and becomes 1/10 when T = 50 ns. As will be described later, it is preferable to set the half width W = 25 ns to be smaller than the delay φ and set the delay time T = 30 to 50 ns. In addition, the pulse energy of the pulse shot 20: 670 mJ, the repetition frequency: 300 Hz, the fourth number of irradiations (10 times, 20 times) is the irradiation of the pulsed laser 20 | According to Fig. 4, Fig. 5 'in general In the conventional method (delay T = 0 ns), there is no process condition (zero is zero) in the number of times of irradiation. If the delay time of the split laser 20 a, 2 0 b is longer, the positive process margin from the delay time T = 20 ns or more is set to the delay time T = 30 ns or more (the intensity of the minimum point is 1/2 or less). Appropriate process conditions (appropriate process margin) can be obtained. Therefore, 20a 2, when, the minimum is the largest, and the wave is ^ 10 to be plotted, and the T-like thunder is caused by the delay. , , the intensity of the -17- (14) 1301293 minimum point 21, that is, the minimum 値 is set to 1/2 or less. However, as can be understood from Fig. 5, when the delay time T = 5 ns or more (the intensity of the minimum point is 1/1 0 or less), the optimum crystallization energy starts to rise sharply, and the intensity of the minimum point 21 is set to 1/10. the above. As shown in Fig. 4, the minimum processing time is less than 1/2 (the number of irradiations is 10 times, about 20 mJ/cm2, and the number of irradiations is 20 times, about 40 mJ/cm2), which is not the same as the above. In the conventional method, the process margin p (zero number of irradiations is zero, and the number of irradiations is 20 times, about 20 mJ/cm 2 ) is improved. The material 12 made of a tantalum film crystallized by the number of times of irradiation in FIG. 4 was observed by SEM (scanning electron microscope), and the size of crystal grains in one field of view was measured to be 90% or more of crystal grains. When the laser wavelength is 308 nm ± 30 nm, the process conditions are used to observe the allowable width (process margin) of the irradiation energy density at which the process conditions can be obtained. As a result, the number of times of irradiation: the process margin of about 10 times (about 20 mJ/cm 2 ) is almost the same as that of the conventional method of φ-dividing lasers, and the number of times of irradiation is 20 times. Crystal particles of almost the same size are obtained. In this way, by setting the intensity of the minimum point to 1/2 or less, it is known that the effect of reducing the number of times of irradiation is obtained, and then the integer N - 3 is formed, and the maximum peak 制作 is made N+1, which satisfies 1/2~ The minimum number of 1/10 is N, and the effect is confirmed by experiment. The pulse waveform 3 of one pulsed laser formed by the overlap of the N + 1 divided lasers is as shown in Fig. 3 'as the starting point of the irradiation -18-(15) 1301293 3〇 The first maximum peak 値4 1 between the first minimum point 3 i , the second maximum peak 値 42 between the first minimum point 3 1 and the second minimum point 32 , the (N-1 )th smallest point 37 and the Nth The Nth maximum peak 値48 between the minimum point 38 and the Nth (N + 1 ) maximum peak 间49 between the Nth smallest point and the end point of the irradiation 50 have a total of N for crystallization of the material 112. The minimum point and the (n + 1) maximal peaks that cause the material 1 1 2 to melt. Then, the crystallization method is a crystallization method in which the intensity of the ηth minimum point is the ηη/ΐο or more and Ιη/2 or less, or the η minimum. The intensity of the point is the crystallization method of the intensity η +1 of the η + 1 maximum peak 之后 after the η + 1 , and is the intensity of the range of In / ίο or more and Ι η / 2 or less. For the material 1 1 2, by 1 The pulsed laser can be melted, crystallized, melted, and crystallized at least sequentially, so that the number of irradiations can be reduced. However, η is an integer ranging from 1 to Ν. For this purpose, the semi-transmission mirrors 103, 104, φ 1〇5, and one full are used in the plural (three on the figure) shown in Figures 7(A), ( Β ), and ( C ). The mirror 106 divides one pulsed laser 102 emitted from the oscillator 101, and forms a complex number of times with a delay time T = 30 ns (the intensity of the minimum point 値 1/2) (four in the figure) The adjacent divided lasers 120a, 120b, 120c, and 120d pass all of the divided lasers 120a, 120b, 120c, and 120d into the common homogenizer 110. Therefore, the delay circuits 107, 108, and 109 are disposed between the respective half mirrors 103 to 105 and the total reflection mirror 106. Thereby, a pulse waveform having a maximum peak N of N+1 and a minimum of N is shown in Fig. 3. The semi-transmissive mirrors 103 to 105, the total reflection mirror 106, and the delay circuits 107 to -19-(16) 1301293 1 09 constitute a divided laser 120a, 120b for delaying the pulsed lasers 102 into plural numbers. , 120c, 120d segmentation means. The intensity In and iη + ι of the two maximum peaks adjacent to the pulse waveform 3 shown in Fig. 3 can be melted by the material 1 1 2 as long as the intensity of the intensity In 2 is less than the minimum point therebetween. There is a maximum peak of maximum intensity among the maximum peaks of the complex intensity In. Actually, by changing the number of the semi-transmissive mirrors 103 to 105 and the delay circuits 1〇7 | to 1 09, a pulse waveform having a maximum peak number of 1 to 7 is irradiated onto the film-shaped tantalum material 1 2, The relationship between the number of peaks of 5, 10, 15 and 20 times of the pulsed laser irradiation and the process margin was measured. The result is shown in Fig. 6. According to Fig. 6, if the maximum number of peaks (and the number of minimum points) is increased, the process margin is expanded, and even if the number of times of irradiation is 5, if the number of peaks is 3 or more, it is known. Good process conditions can be obtained. This number of illuminations is the number of pulsed lasers 102. The number of times the pulsed laser of each material 12 is irradiated can be sufficiently less than 1 〇 φ times. [Embodiment] A first embodiment will be specifically described with reference to Fig. 7. The pulsed laser 102 of the pulse waveform emitted by the XeCl laser laser oscillator 101 with a pulse energy of 670 mJ and a repetition frequency of 300 Hz is provided by four mirrors 1 having reflectances of R1, R2, R3, and R4. 〇3, 104, 105, 106 are divided and converted by direction, and the adjacent circuits are separated by delay circuits 107, 108, 1〇9 -20-(17) 1301293 provided between the mirrors 103, 104, 105, and 106. The lasers 120a to 120d are time-delayed with each other, and are sequentially injected into the long-axis homogenizer 1 10, and shaped into a long axis by a known means: a spatial intensity distribution of a rectangular wave of 200 mm length, and a film irradiated on the glass of the material 12 A-Si film II2 having a thickness of 50 nm. The reflectance of the mirror 103 is R1, the reflectance of the mirror 104 is R2, the reflectance of the mirror 105 is R3, and the reflectance of the total reflection mirror 106 is R4 = 100%. At this time, the aforementioned reflectance is set to: Rl=25%, R2=33.3%, R3=50%, R4=l〇〇%, to obtain a pulse waveform which is a maximum peak, and the intensity can be generated in a period of 30 ns to generate the third of four. Figure 3 of the waveform. The laser laser emits a non-linearly polarized light in the ultraviolet region where the absorption of the a-Si film 12 is high, and a pulsed light (wavelength 3 0 8 nm) can be obtained. In addition, each delay circuit 107, 108, The 109 series includes a total number of total reflection mirrors required to generate an optical path difference, and the total reflection mirrors are used to set a time difference between the divided lasers 120a to 120d. In order to perform an individual delay time: a delay of 30 ns, a light path of 9 m is set. . In the direction orthogonal to the long axis of the pulsed laser 1 〇 2, a short-axis homogenizer 1 1 1 as shown in Fig. 7 (C) is set, and a space of a rectangular wave having a short axis width of 0.4 mm is shaped. The intensity distribution is irradiated to the material 1 12 by the incident angle α = 5 ° of the incident light to the normal of the material 1 1 2 . Further, the a-Si film 112 on the glass is placed on a stage (not shown) and scanned at a speed of V (mm/s) in the short-axis direction. Then, according to V = 0.4 · 300 / Z, set V = 24, 12, 8, 6 mm / s, to make the number of irradiations at one of the a - S i films 1 1 2: Z = 5, 10, 15, 20 times. The homogenizers 110 and 111 constitute a shaping means for shaping the intensity distribution of a plurality of -21 - (18) l3 〇 1293 divided lasers so as to be superimposed on the material 1 1 2 . The material 112 of the crystallized ruthenium film after the irradiation of the pulsed laser 102 is observed by SEM (scanning electron microscope), and the size of the crystal grains in one field of view is measured, and 90% or more of the crystal grains are used. When the laser wavelength is 308 nm ± 30 nm, the process conditions are used to measure the allowable width (process margin) of the irradiation energy density at which the process conditions can be obtained. As a result, the number of times of the illuminating: 5, 10, 15, and 20 times is as shown in Fig. 6 as the maximum peak number 4, about 20, 40, 60, 100 mJ/cm2, even in the past. The number of times the process is not feasible: 5 times or less, can also be obtained as a condition of the process. By setting the incident angle α of the pulsed laser 102 (divided laser 120 a to 120d) to the material 1 1 2 to be equal to or greater than the vertical line (normal), the crystal grain ratio can be made to the laser 1 The wavelength of 02 is larger. Therefore, the incident angle α of the incident ray to the normal of the material 1 1 2 is set to be Γ or more. Further, the laser I repetition frequency is a relationship of several hundred Hz, and the pulsed laser 102 is irradiated at intervals of several ms. Therefore, by irradiation of a pulsed laser of a plurality of pulses, ruthenium (material 1 1 2 ) is repeatedly melted and cooled and crystallized, and the wavelength of the laser beam to be irradiated is almost the same or more uniform, so that crystallization can be performed. The difference in the size of the particles becomes smaller. When one pulsed laser beam 102 is irradiated as a plurality of divided lasers 120a to 120d, it is considered that a pulse of a plurality of pulses corresponding to the number of divisions can be obtained by one pulsed laser beam 102. The same effect as a laser. In addition, as described in Japanese Patent Laid-Open No. 2004-172424, a pulsed laser 〇2 of a -22-(19) 1301293 material 1 1 2 is irradiated, such as a non-linearly polarized laser, which can be used without a polarizing plate. If the incident angle of the incident light to the normal of the material is Γ or more, the crystal grain may be equal to the wavelength of the laser beam 2 by repeated irradiation of the pulsed laser 102 or It is larger than the wavelength of the laser. A laser laser is a random polarized light that does not have a specific polarized light. By setting the incident angle of the incident light to the normal of the material to be Γ or more, a large crystal grain can be obtained. BRIEF DESCRIPTION OF THE DRAWINGS Fig. 1 is a schematic view showing a crystallization apparatus for a film material according to an embodiment of the present invention. Fig. 2 is a view similarly showing the time on the horizontal axis and the intensity on the vertical axis. The cut laser, the second (a) graph shows the line diagram of the split laser, and the second (b) diagram shows the split mine. A line graph of overlapping states of shots. Fig. 3 is a line diagram similarly showing the overlapping state of the laser beam divided by the horizontal axis time and the vertical axis intensity. Fig. 4 is a line diagram similarly showing the delay time-process margin characteristic. Fig. 5 is a line diagram similarly showing the delay time-optimal crystallization energy characteristic. Fig. 6 is a line diagram showing the maximum peak number-process margin characteristic in the same manner. Fig. 7 is a view showing the crystallization apparatus of the film material used in the first embodiment in the same manner, and Fig. 7(A) shows the laser oscillator and the division means. 23-(20) 1301293 槪 thumbnail '弟7 (B) The figure shows a sketch of the material omitted from the chemical, and a sketch of the material of the seventh (C) material. [Symbol description of main components] 1, 1 0 1 : laser oscillator, 2, 3: pulse waveform, 1 2, 1 1 2: film-like tantalum material, 20, 102: pulsed laser, 20a, 20b, 120a, 120b, 120c, 21, 31, 32, 37, 38: minimum point, 22, 23, 41, 42, 48, 49: @ -) 3 0 : starting point, 5 0 : ending point, 103, 104, 105: semi-transmissive mirror 106: total reflection mirror (dividing means), 107, 108, 109: delay circuit (divided into 1 1 0, 1 1 1 : homogenizer (shaping means) 221 : semi-transmissive mirror ( Dividing means 222: total reflection mirror (dividing means), 223: delay circuit (dividing means). Shaft homogenizer, which is represented by the long axis, from the short-axis homogenizer 1 2 0 d: split laser, C-peak , (dividing means), cutting means), ), -24 -