201132716 六、發明說明： 【發明所屬之技術領域】 本發明係關於在半導體之製程中用於使氧化矽質膜形 成之被覆組成物者。更具體而言，係關於在半導體之製程 作爲絕緣膜使用之用於使氧化矽質膜形成之含聚矽氮烷被 覆組成物。 【先前技術】 近來，於半導體裝置係開始要求更高的積體密度，同 時可相應之製造技術正被改良。並且，在該等半導體裝置 的製造過程之一的使絕緣膜形成之步驟，係變得有必要埋 設窄隙。 要埋設如此之窄隙，已知使用含全氫聚矽氮烷被覆組 成物。全氫聚矽氮烷係基本骨架由Si-N ' Si-H、N-H鍵所 構成之聚合物，具有藉由在含氧及/或水蒸氣之氣體環境 燒製而Si-N鍵被Si-O鍵取代，得到純度高之氧化矽質膜 這樣的特徵。 然而，隨著半導體所被要求之積體密度變得更高，間 隙開始變得更窄。以往所知之全氫含聚矽氮烷被覆組成物 —般而言，雖道是埋設性優良，但爲了要達成近來所要求 之高積體密度，係開始有改良之必要。具體而言，在以往 之被覆組成物，要使埋設性與塗布性兼顧開始變得困難。 如此問題的原因之一，已知有全氫聚矽氮烷的分子量 分布。例如，於專利文獻1係揭示有使用重量平均分子量 爲4000至8000，重量平均分子量及數量平均分子量之比 201132716 爲3.0至4.0之全氫聚矽氮烷的旋轉塗布玻璃（spino 組成物。此外，於專利文獻2，係揭示有重量平均 爲3000至6000之含聚矽氮烷旋轉塗布玻璃。再者 利文獻3係揭示有聚苯乙烯換算分子量爲700以下 氮烷量爲總聚矽氮烷量之1 〇%以下的氧化矽系被覆 用塗布液。該等係皆爲欲控制聚矽氮烷之分子量分 良塗布性等者。 根據本發明人等之探討，若使用重量平均分子 全氫聚矽氮烷則埋設性有提升之傾向，但塗布時變 產生條紋，相對地若使用重量平均分子量大的全氫 烷，則條紋的產生受抑制且塗布性改良，但埋設性 之傾向。結果未充分埋設至窄隙的深部，容易發生 燒製使氧化矽質膜形成時，間隙深部的氫氟酸所致 率變大這樣的問題而有如此之問題點。如此之問題 僅有專利文獻1至3所記載之分子量分布之控制時 分，期望更進一步的改良。 [先前技術文獻] [專利文獻] [專利文獻1]日本特開2001-319927號說明書 [專利文獻2]日本特開2005-150702號說明書 [專利文獻3]日本特開平8 -2 693 99號說明書 [專利文獻4]日本專利第1474685號說明書 [專利文獻5 ]日本專利第2 6 1 3 7 8 7號說明書 n glass) 分子量 ，於專 之聚矽 膜形成 布，改 量小的 得谷易 聚矽氮 有劣化 塗布後 之蝕刻 點係在 則不充 201132716 【發明内容】 [發明欲解決之課題] 如上所述，以往之被覆組成物係在近來所要求之於欲 對具有窄隙之基板使氧化矽質膜形成時，無法充分地兼顧 埋設性與塗布性。本發明之目的係有鑒於如此之問題點， 而提供可充分地埋設窄隙，換言之爲縱橫比大的間隙、且 塗布時不產生條紋，用於使半導體裝置之氧化矽質膜形成 的被覆組成物。 [爲解決課題之手段] 本發明所致之被覆組成.物，其爲含全氫聚矽氮烷與溶 劑而成之被覆組成物，其特徵爲該全氫聚矽氮烷之分子量 分布曲線在分子量800至2,500之範圍、與分子量3,000 至8，000之範圍各具有極大値，且重量平均分子量Mw與 數量平均分子量Μη之比Mw/Mn爲6至1 2。 此外，本發明所致之氧化矽質膜之形成方法，其特徵 爲含有以下步驟：於具有凹凸之基板的表面上塗布被覆組 成物之塗布步驟，該被覆組成物爲含全氫聚矽氮烷與溶劑 而成之被覆組成物，其中該全氫聚矽氮烷之分子量分布曲 線在分子量800至2,500之範圍、與分子量3,000至8,000 之範圍各具有極大値，且重量平均分子量Mw與數量平均 分子量Μη之比Mw/Mn爲6至12;及將完成塗布之基板在 小於1 〇〇〇 °C之氧氣環境或含水蒸氣之氧化氣體環境進行加 熱處理使前述組成物轉化爲二氧化矽膜之硬化步驟。 [發明的效果] 201132716 根據本發明之被覆組成物，可兼顧含聚矽氮烷化合物 被覆組成物之塗布性與埋設性，更且亦可改善所得之氧化 矽質膜的膜物性。 【實施方式】 [實施發明之形態] 以下，就本發明的實施形態詳細地説明。 被覆組成物 本發明所致之被覆組成物，係含全氫聚矽氮烷與可將 其全氫聚矽氮烷溶解之溶劑而成。 用於本發明之全氫聚矽氮烷，係如後述一般需要具有 特定之分子量及分子量分布，但其結構並無特別限定，只 要不損及本發明之效果可選擇任意結構者。爲無機化合物 之全氫聚矽氮烷，係具有僅由矽、氮、及氫構成，藉由燒 製而使氧化矽質膜形成時不純物不易混入之特徵。如此之 全氫聚矽氮烷的具體結構係可以下述一般式（I)表示。 -(SiH2-NH)„- (I) 式中，η係表示聚合度之數。 另外在不損及本發明之效果的範圍，亦可少量含有（I) 式之氫的部分或全部被取代成烷基、烯基、環烷基、芳基、 烷矽基、烷胺基或烷氧基等之聚矽氮烷化合物。 本發明所致之聚矽氮烷組成物，雖係含前述之全氫聚 矽氮烷而成者，但其全氫聚矽氮烷之分子量分布曲線係在 分子量800至2,500之區域、與分子量3,000至8，000之區 域具有極大値者。此時，於這兩個極大値之間係存在一個 201132716 以上，較佳爲一個的極小値。 具有如此之分子量分布曲線的全氫聚矽氮烷，雖係可 用任意方法進行調製’但最簡單係可藉由混合分子量相對 地大的全氫聚矽氮烷、與分子量相對地小的全氫聚矽氮院 而獲得。更具體地說，較佳爲藉由將重量平均分子量800 至2,500，尤其1，000至2，200的全氫聚砂氮院（以下，爲 簡單化以低分子量聚矽氮烷稱之）、與重量平均分子量 3,000至8,000，尤其3,500至7,000的全氫聚矽氮烷（以 下’爲簡單化以高分子量聚矽氮烷稱之）混合而獲得。此 係雖無特別限定混合前的全氫聚矽氮烷之合成方法，但可 藉由例如專利文獻4或5所記載之方法而合成。 不僅全氫聚矽氮烷，高分子化合物由於在分子量分布 具有範圍，在將分子量不同的兩種高分子化合物混合時、 混合前後，有時分子量分布的極大値之位置會變化。此係 尤其在兩種高分子化合物之分子量分布的極大値分子量相 近時容易發生，藉由混合，具極大値的分子量有接近的傾 向。視情形極大値亦可能變成一個。然而，將具有如前所 述之重量平均分子量的兩種全氫聚砍氮院混合時，因分子 量差爲大，一般而言極大値係不變成一個。此外，在本案 發明中，由於認爲藉由使具有二個極大値之間的分子量之 成分減少而展現本發明的效果，故有必要至少混合兩種全 氫聚矽氮烷使兩個極大値之間具有極小値。 本發明中，爲了達成目的分子量分布而混合兩種全氫 聚矽氮烷時，各全氫聚矽氮烷的分子量分布係以窄者爲 201132716 佳。此係因所混合之任一者、或兩者的全氫聚矽氮烷之分 子量分布若廣，則分布曲線的兩個極大値之間不易顯出極 小値，此外本發明的效果亦有變小之傾向之故。具體而言， 就進行混合前的兩種全氫聚 量Mw與數量平均分子量Μη 要使全氫聚矽氮烷的分 矽氮烷所含之高分子量成分 係爲簡便。以如此之除去高 成分的簡單方法而言，舉出 方法。亦即，全氫聚矽氮烷 解度越低、分子量越小溶解 解度差，使該全氫聚矽氮烷 氮烷之一部分的程度的溶解 性成分，而可分別成當作不 量成分、與溶解於溶劑中之 被過濾分離之不溶性成分則 溶解之成分則係除去低分子 在此，全氫聚矽氮烷之 因此以某溶劑除去高分子量 的另一溶劑除去低分子量， 窄。在如此之方法，雖常有 或低分子量成分之情況，但 聚合度的化合物之混合物即 窄，係簡便且有效的方法。Β 矽氮烷之各者，重量平均分子 之比Mw/Mn較佳爲1.1至1.8。 子量分布狹窄，由除去全氫聚 、及/或低分子量成分而進行 分子量成分、及/或低分子量 利用溶解度的分子量依存性之 ，一般而言係有分子量越大溶 度越高之傾向。利用如此之溶 溶解於具有可溶解該全氫聚矽 性之溶劑，藉由過濾分離不溶 溶性成分被過濾分離之高分子 低分子量成分。亦即，若除去 係除去高分子量成分，若除去 量成分。 溶解性係視使用之溶劑而異， 成分後，亦可使用溶解性不同 進一步使分子量分布的寬度狹 不能完全地除去高分子量成分 爲了使不同分子量，亦即不同 高分子化合物的分子量分布狹 I用於如此之用途的溶劑而言， 201132716 例如以碳氫化合物爲合適。例如若爲院類，隨著碳數變多 有可溶解分子量更大的全氫聚矽氮烷之傾向。一般而言， 可使用碳數5至10左右的碳氫化合物。 此外’爲使高分子化合物之分子量分布狹窄，亦可藉 由一般所用之層析法等而分離成全氫聚矽氮烷之各分子 量。但是，若使用層析法則處理時間可能會變長，從生產 效率的觀點看來以利用對前述溶劑之溶解性差的方法爲 佳。 還有’不僅進行用於使全氫聚矽氮烷之分子量分布狹 窄的處理，藉由進行合成方法或合成原料的調整而合成分 子量分布窄的全氫聚矽氮烷亦爲有效。 另外，在混合兩種全氫聚矽氮烷前，於除去各全氫聚 矽氮烷所含之高分子量成分或低分子量成分時，較佳爲將 低分子量全氫聚砍氮院之局分子量成分，又，高分子量全 氫聚矽氮烷之低分子量成分除去。藉由如此地減少對應於 分子量分布曲線的兩個極大値之中間區域之成分，更強烈 地展現本發明之效果。 以如上之方法準備具有不同分子量之全氫聚矽氮烷， 混合該等時，其混合比較佳爲低分子量聚矽氮烷與高分子 量聚矽氮烷之重量比爲3:7至6:4，更佳爲4:6至6:4。混 合比若在此範圍外，則有時塗布性與埋設性之平衡會因而 變差。 本發明中具有特定分子量分布之全氫聚矽氮烷，係如 前所述一般，以藉由混合分子量不同的兩種全氫聚矽氮烷 -10- 201132716 而獲得爲簡便，但亦可藉由其他方法而獲得。例如，藉由 準備具有較廣之分子量分布的全氣聚砂氮院，然後以層析 法除去分子量在2,500至3，000附近之中間區域成分，而可 達成期望之分子量分布。 此外，用於本發明之全氫聚矽氮烷係需要其重量平均 分子量Mw與數量平均分子量Μη之比Mw/Mn爲6至12, 較佳爲7至1 0。如此之比Mw/Mn係如前所述一般，在混 合兩種全氫聚矽氮烷的情形，在將低分子量全氫聚矽氮 烷、與高分子量全氫聚矽氮烷以3:7至6:4之重量比混合 時可達成。 本發明所致之被覆組成物，係含有前述之可溶解全氫 聚矽氮烷的溶劑而成。以如此之溶劑而言，只要係可溶解 前述之各成分者則無特別限定，但以較佳之溶劑的具體例 而言，舉出下列者： (a)芳香族化合物，例如苯、甲苯、二甲苯、乙苯、二 乙苯、三甲苯、三乙苯等，（b)飽和烴化合物，例如正戊烷、 異戊烷、正己烷、異己烷、正庚烷、異庚烷、正辛烷、異 辛烷、正壬烷、異壬烷、正癸烷、異癸烷等，（c)脂環式 烴化合物，例如乙基環己烷、甲基環己烷、環己烷.、環己 烯、對薄荷烷、十氫萘、雙戊烯、檸檬烯等，（d)醚類’例 如二丙醚、二丁醚、二***、甲基三級丁基醚、苯甲醚等’ 及（e)酮類、例如甲基異丁基酮）等。該等之中，更佳爲（b) 飽和烴化合物、（c)脂環式烴化合物（d)醚類、及（Ο酮類。 該等之溶劑，爲調整溶劑的蒸發速度、爲降低對人體 -11 - 201132716 之有害性、或爲調製各成分之溶解性，而係亦可使用經混 合適當之兩種以上者。 用於本發明之被覆組成物，係視需要亦可含有其他之 添加劑成分。如此之成分，可舉出例如黏度調整劑、交聯 促進劑等。此外，以用於半導體裝置時，鈉之聚集（gathering) 效果等爲目的，亦可含有磷化合物，例如參（三甲基矽基） 磷酸酯等。 此外，前述之各成分的含有量，雖係視目的組成物的 用途而改變，但全氫聚矽氮烷的含有率較佳爲10至25重 量％，更佳爲設爲12至22重量％。一般而言全氫聚矽氮烷 的含量若過高則被覆組成物的黏度變高、埋設性或塗布性 有劣化之傾向，又若過低則所形成之氧化矽質膜的厚度有 不足之傾向。 氧化矽質膜之製造法 根據本發明之氧化矽質膜之製造法，可於具有溝或孔 洞等之縫隙的基板上，形成充分地埋設至縫隙之深部、膜 面平坦、膜質亦均勻的被覆膜。因此，電子裝置的電晶體 部分或電容器部分的作爲平坦化絕緣膜（前金屬（pre-metal) 絕緣膜）而形成，再於帶有溝之基板上使氧化矽質膜形成 並將溝埋封，亦可形成槽隔離（trench isolation)結構。以下 基於使槽隔離結構形成之方法說明本發明。 (A)塗布步驟 本發明所致之被覆組成物，係適於形成基板上的槽隔 離結構者。於使槽隔離結構形成時，準備具有期望之溝圖 -12- 201132716 案的矽等之基板。要形成此溝’雖可使用任意方 由例如以下所示之方法可使其形成。 首先，於矽基板表面藉由例如熱氧化法使 膜。在此，使其形成之二氧化矽膜的厚度一般爲5 視需要於所形成之二氧化矽膜上藉由例如減 使氮化矽膜形成。該氮化矽膜係可作爲之後的蝕 之光罩、或後述之硏磨步驟中之終止層而使其 者。氮化矽膜，於使其形成時，一般係以1 〇〇至 厚度使其形成。 於如此使其形成之二氧化矽膜或氮化矽膜之 阻。視需要使光阻膜乾燥或硬化後，以期望的圖 光及顯影使圖案形成。曝光的方法係可以光罩曝 曝光等任意的方法來進行。此外，光阻從解析度 來看，亦可選擇任意者使用。 將所形成之光阻膜當作光罩，依序蝕刻氮化 其下之二氧化矽膜。藉此操作而於氮化矽膜及二 形成期望之圖案。 將經形成圖案之氮化矽膜及二氧化矽膜作爲 蝕刻矽基板，使槽隔離溝形成。 所形成之槽隔離溝的寬度，係由將光阻膜曝 所決定。半導體元件中槽隔離溝雖係視目的之半 而不同，但寬度係一般爲0.02至10 μηι，較佳爲 μιη，深度係200至1000 nm，較佳爲300至700 明所致之方法，與以往之槽隔離結構的形成方法 法，但藉 二氧化矽 至 3 0 nmo 壓CVD法 刻步驟中 發揮功能 400 nm 的 上塗布光 案進行曝 光、掃描 等之觀點 矽膜及在 氧化矽膜 光罩，乾 光的圖案 導體元件 〇.〇5 至 5 nm。本發 相比，由 -13- 201132716 於可均勻地埋設至更窄、更深的部分，故係適於使更窄、 更深的槽隔離結構形成的情況者。尤其，在以往之氧化矽 質膜形成用組成物或氧化矽質膜的形成方法，係難以形成 均勻的氧化矽質膜至溝的深部分，溝的寬度一般爲0.5 μπι 以下，尤其〇 . 1 μιη以下，於形成縱橫比爲5以上之槽隔離 結構時，藉由使用本發明所致之氧化矽質膜形成用組成物 而可使其均一地形成溝內的氧化矽質膜。 接著，於如此所準備之矽基板上，使氧化矽質膜的材 料所成之前述被覆組成物的塗膜形成。 被覆組成物係可以任意的方法塗布於基板上。具體而 言，舉出旋轉塗布、簾式塗布、浸漬塗布、及其他。該等 之中，從塗膜面的均一性等之觀點來看特佳爲旋轉塗布。 爲兼顧氧化矽質膜形成用組成物塗布後之槽溝埋設性 及塗布性，所塗布之塗膜的厚度一般爲10至l,000nm，較 佳爲50至800 nm。 塗布的條件雖視組成物的濃度、溶劑、或塗布方法等 而改變，但舉出以旋轉塗布爲例則係如以下所述。 最近係爲了改善製造之產率，常於大型基板使元件形 成，但爲了要於8吋以上的矽基板均一地使氧化矽質膜形 成用組成物的塗膜形成，組合多個階段的旋轉塗布係爲有 效。 首先，於矽基板的中心部、或基板整面，以平均地形 成塗膜的方式，於含中心部之數處，一般係每一塊矽基板 滴下0.5至20cc的組成物。 -14- 201132716 接著，爲了將經滴下之組成物擴展至矽基板整面，以 較低速且短時間，例如旋轉速度50至5 00r pm、0.5至10 秒鐘而使其旋轉（預旋轉）。 接著，爲了使塗膜成爲期望之厚度，以較高速，例如 旋轉速度5 00至4500rpm、0.5至8 00秒鐘而使其旋轉（主 旋轉）。 再者，爲了減低在矽基板之周邊部分的塗膜之遽增、 且將塗膜中之溶劑盡可能地乾燥，以相對於前述主旋轉之 旋轉速度快5 00rPin以上之旋轉速度，例如旋轉速度1000 至5000 rpm、5至300秒鐘而使其旋轉（終旋轉）。 該等之塗布條件係視使用之基板的大小、或目的之半 導體素子的性能等而適當調整。 (B)硬化步驟 塗布被覆組成物後，可視需要付諸預烘焙步驟。在預 烘焙步驟，係以塗膜中所含溶劑之完全除去、與塗膜之前 硬化爲目的者。尤其在使用含聚矽氮烷組成物之本發明的 氧化矽質膜之形成方法中，係藉由進行預烘焙處理，提升 所形成之氧化矽質膜的緻密性，因此較佳爲組合預烘焙步 驟。 通常，在預烘焙步驟係於實質上採取以固定溫度進行 加熱之方法。此外，於硬化時係爲了防止塗膜收縮、縫隙 部分變成凹痕、於縫隙內部產生空隙，而較佳爲控制預烘 焙步驟中之溫度、經時地使其上升同時進行預烘焙。預烘 焙步驟中之溫度通常係在50 °C至400 °C，較佳爲在1〇〇至 -15- 201132716 3 0 (TC之範圍內。預烘焙步驟之需要的時間一般係10秒鐘 至30分鐘，較佳爲30秒鐘至10分鐘。 要使預烘焙步驟中之溫度經時地上升，舉出使基板所 被放置之氣體環境的溫度階段性地上升之方法、或使溫度 單調增加地上升之方法。在此，預烘焙步驟中之最高預烘 焙溫度，由從被覆膜除去溶劑之觀點來看，一般設定成比 用於氧化矽質膜形成用組成物之溶劑的沸點更高的溫度。 另外，本發明所致之方法中，於組合預烘焙步驟時， 將變得比預烘焙更高溫的基板降低溫度前，以將較佳爲 5 〇 °C以上、預烘焙時之最高溫度以下的溫度之基板付諸硬 化步驟爲佳。由將降低溫度前之基板付諸硬化步驟，可節 省再度使溫度上升之能量與時間。 接著，爲了使含聚矽氮烷塗膜轉化成氧化矽質膜使其 硬化，將基板整體進行加熱，付諸硬化步驟。通常，一般 係將基板整體投入硬化爐等並進行加熱。 硬化係較佳爲使用硬化爐或加熱板，在含水蒸氣之惰 性氣體或氧氣體環境下進行。水蒸氣於使聚矽氮烷充分轉 化成氧化矽質膜（亦即二氧化矽）係爲重要，較佳爲設爲 3 0 %以上、更佳爲50%以上、最佳爲70%以上。尤其水蒸氣 濃度若爲80%以上，則有機化合物對氧化矽質膜之轉化變 得容易進行，空隙等缺陷產生變少，氧化矽質膜之特性改 良故佳。於使用惰性氣體作爲氣體環境氣體時，使用氮、 氬、或氦等。 爲使其硬化之溫度條件，係視使用之氧化矽質膜形成 -16- 201132716 用組成物的種類、或步驟的組合方式而改變。然而 高者，聚矽氮烷轉化成氧化矽質膜之速度有變快之 此外，溫度低者，矽基板之氧化或結晶結構之改變 對裝置特性之不良影響有變小之傾向。由如此之 言，在本發明中之硬化步驟，係通常以1 00 0 °c以下 爲400至700°c進行加熱。在此，至目標溫度之昇 一般係1至100 °C /分鐘，到達目標溫度之後的硬化 般係1分鐘至1 〇小時，較佳爲1 5分鐘至3小時。 亦可使硬化溫度或硬化氣體環境的組成階段性地改 由此加熱，聚矽氮烷轉化成二氧化矽而成爲氧化矽 本發明所致之氧化矽質膜的形成方法，雖係以 各步驟爲必要者，但視需要亦可組合硏磨步驟或蝕 等之進一步的步驟。 若使用各例說明本發明則係如以下所述。 含成例1 低分子量聚矽氮烷之合成 將400g純度99 %以上之二氯矽烷一邊攪拌一邊 5 k g 0 °C的無水吡啶。將此混合物的溫度維持於0 °C 一邊攪拌1.22kg純度99.9%的氨氣一邊注入至混合 一邊將混合物的溫度維持於〇°C —邊繼續攪拌: 小時反應。將乾燥氮吹入反應後之混合物30分鐘， 剩的氨，其後從漿體狀之反應混合物過濾分離氯化 到濾液A。將二甲苯混合至所得之濾液A中加熱至 在20mmHg之減壓下進行蒸餾除去吡啶，作成含重 分子量1450的聚合物之20重量％濃度之溶液。 ，溫度 傾向， 所致之 觀點而 ，較佳 溫時間 時間一 視需要 變。藉 質膜。 前述之 刻步驟 注入至 ，同時 物。 隹行1 2 除去過 銨，得 5 0°C， 量平均 -17- 201132716 將所得之 20重量％的二甲苯溶液加熱至 50°C，在 lOmmHg之減壓下進行蒸餾除去二甲苯。將正戊院力卩入至 所得之無色透明液體中，作成1 〇重量％濃度的白色溶液。 將此溶液以過濾精度〇·2 μιη的過濾器進行過濾得到聚合物 溶液。將二丁醚混合於此聚合物溶液且加熱至50°C，在 2 OmmHg之減壓下進行蒸餾除去正戊烷，作成含重量平均 分子量1100、重量平均分子量Mw與數量平均分子量Μη 之比Mw/Mn爲1.45之聚合物的20重量％濃度之聚合物溶 液1。 合成例2 高分子量聚砂氮垸之合成 與合成例1同樣地進行而調製濾液A，進一步在密閉 系統以1 5 0 °C加熱3小時。冷卻至室溫後，回到常壓，將 二甲苯混合至所得之溶液並加熱至50°C，在20 mmHg之減 壓下進行蒸餾除去吡啶，並作成含重量平均分子量6000之 聚合物的20重量％濃度之溶液。 將所得之20重量％的二甲苯溶液加熱至50°C，並在 10 mmHg之減壓下進行蒸餾除去二甲苯。加入正庚烷至所 得之白色粉末，作成1 〇重量％濃度之分散液。使用玻璃過 濾器（ADVANTEC東洋.股份有限公司製：GF-75 (商品名）） 減壓過濾該分散液，並除去溶劑。將所得之白色粉末溶解 於二丁醚，作成含重量平均分子量爲64〇0、重量平均分子 量Mw與數量平均分子量Μη之比Mw/Mn爲1.22之聚合物 的20重量％濃度之聚合物溶液2。 合成例3 超高分子量聚矽氮烷之合成 -18 - 201132716 與合成例1同樣地進行而調製濾液A，在密閉系統以 1 5 0 °c加熱6小時。冷卻至室溫後，回到常壓，將二丁醚混 合至所得之溶液並加熱至5(TC，在20 mmHg之減壓下進行 蒸餾除去吡啶，並作成含重量平均分子量9200之聚合物的 2〇重量％濃度之聚合物溶液3。 實施例1 混合60g之聚合物溶液1與40g之聚合物溶液2。混 合後之聚合物溶液，於分子量分布曲線之分子量爲6300之 位置與650之位置具有極大値，又Mw/Mn爲10。 準備形成有深度0.5 μιη且具有0.05、0.1、0.2、及0.5 μπι之寬度的溝、表面被氮化矽內襯層所被覆之基板作爲矽 墓板。於此基板藉由旋轉塗布將所調製之聚合物溶液塗布 作爲被覆組成物。塗布條件係設爲，預旋轉：3 00rpm/5秒 鐘、主旋轉：1 000rpm/20秒鐘、終旋轉：1500rpm/10秒鐘。 觀察塗布後之膜面後，可確認沒有產生條紋、可達成優良 之塗布性。 進一步將此塗布後之基板在加熱板上150 °C預烘焙3 分鐘，繼續不使其冷卻而導入至純氧氣體環境下之燒製 爐。在燒製爐內以昇溫速度10 °C/分鐘至加熱800 °C，進一 步在水蒸氣濃度80%之氧氣體環境下燒製30分鐘。測定所 得之燒製膜的FT-IR後，觀測到歸屬於Si-Ο鍵之波數1080 cnT 1之吸收，確認得到氧化矽質膜。另一方面，觀測不到 歸屬於N-H鍵及Si-H鍵之波數3380 cm·1及2200 cm·1之 吸收，確認全氫聚矽氮烷已轉化成氧化矽。 -19- 201132716 此外，使用含0.5重量％之氫氟酸與30重量％之氟化 銨的水溶液作爲蝕刻溶液，在2 3 °C進行蝕刻而測定對於熱 氧化氧化矽膜之相對蝕刻率後，爲1 ·48。 將燒製後之基板以相對溝之長度方向爲直角之方向切 斷後，浸漬於含0.5重量％之氫氟酸與30重量％之氟化銨 的水溶液30秒鐘，進一步用純水充分地洗淨而後使其乾 燥。 將基板之截面，藉由掃描型電子顯微鏡，以50,000倍 而從痢直於截面之方向的仰角30°上方觀察各溝之最深 部，評估蝕刻量。即使溝寬度改變，鈾刻量之改變亦僅有 一點點，可確認即使在寬0.05 μηι之溝的最深部中亦充分 地形成有緻密的氧化矽質膜。 實施例2 混合40g之聚合物溶液1與60g之聚合物溶液2。混 合後之聚合物溶液，係於分子量分布曲線之分子量爲6250 的位置與680之位置具有極大値，又Mw/Mn爲10。 將所調製之聚合物溶液當作被覆組成物，與實施例1 同樣地進行，並塗布於矽基板上。觀察塗布後之膜面後， 可確認沒有產生條紋、可達成優良之塗布性。 進一步將此塗布後之基板與實施例1同樣地進行燒 製。測定所得之燒製膜的FT-IR後，觀測到歸屬於Si-Ο鍵 之波數1 080 cnT1之吸收，確認得到氧化矽質膜。另一方 面，觀測不到歸屬於N-H鍵及Si-H鍵之波數3 3 8 0 cm·1及 2 2 00 cnT1之吸收，確認全氫聚矽氮烷已轉化成氧化矽。此 -20- 201132716 外，與實施例1同樣地測定對於熱氧化氧化矽膜之相胃 刻率後，爲1 . 5 0。 將燒製後之基板的截面藉由與實施例1同樣的方 觀察，並評估蝕刻量。即使溝寬度改變，蝕刻量之改變# 僅有一點點，可確認即使在寬0.05 μηι之溝的最深部中亦 充分地形成有緻密的氧化矽質膜。 比較例1 將聚合物溶液1當作被覆組成物，與實施例1同樣地 進行，並塗布於矽基板上。觀察塗布後之膜面後’確認從 中心部分朝向周邊部分產生複數條紋，同時塗布性不充分。 進一步將此塗布後之基板以與實施例1同樣的方法燒 製，並以電子顯微鏡觀察截面。在寬〇. 1 μιη以上之溝，係 蝕刻量之改變僅有一點點，可確認即使在最深部中亦充分 地形成有緻密的氧化矽質膜。但是，在寬〇.〇5 μηι之溝的 最深部中，蝕刻量大，確認於該部分係未形成緻密的氧化 矽質膜。 比較例2 將聚合物溶液2當作被覆組成物，與實施例〗同樣地 進行，並塗布於矽基板上。觀察塗布後之膜面後，確認沒 有產生條紋，可達成優良之塗布性。 進一步將此塗布後之基板，以與實施例1同樣的方法 進行燒製，以電子顯微鏡觀察截面。在寬0·2μιη以上之溝， 係蝕刻量之改變僅有一點點，可確認即使在最深部中亦充 分地形成有緻密的氧化矽質膜。然而，在寬0.05μηι與〇.】 -21- 201132716 μπι之溝的最深部中，係蝕刻量大，可確認於該部分係未形 成緻密的氧化矽質膜。 比較例3 · 將聚合物溶液3當作被覆組成物，與實施例1同樣地 進行，並塗布於矽基板上。觀察塗布後之膜面後，可確認 沒有產生條紋、可達成優良之塗布性。 進一步將此塗布後之基板以與實施例1同樣的方法燒 製，並以電子顯微鏡觀察截面。在寬0.2μιη以下之溝，係 辨識出空隙，可確認埋設性有改良的空間》 【圖式簡單說明】 ^ΙΊΓ 無。 【主要元件符號說明】 4nt 挑。 •22-201132716 VI. Description of the Invention: TECHNICAL FIELD The present invention relates to a coating composition for forming a cerium oxide film in a semiconductor process. More specifically, it relates to a polyazoxide-containing coating composition for forming a cerium oxide film which is used as an insulating film in a semiconductor process. [Prior Art] Recently, a higher integrated density has been demanded in semiconductor devices, and corresponding manufacturing techniques are being improved. Further, in the step of forming an insulating film in one of the manufacturing processes of the semiconductor devices, it is necessary to embed a narrow gap. To embed such a narrow gap, it is known to use a perhydropolyazane-containing coating composition. A perhydropolyazane-based polymer having a basic skeleton consisting of Si-N 'Si-H and NH bonds, having Si-N bonds by Si- in a gas atmosphere containing oxygen and/or water vapor. The O bond is substituted to obtain a characteristic of a ruthenium oxide film having a high purity. However, as the bulk density required for semiconductors becomes higher, the gap begins to become narrower. The perhydrogen-containing polyazane-coated composition known in the prior art is generally excellent in embedding property, but it is necessary to improve it in order to achieve a high density which has recently been required. Specifically, in the conventional coated composition, it is difficult to achieve both the embedding property and the coating property. One of the causes of such a problem is known to have a molecular weight distribution of perhydropolyazane. For example, Patent Document 1 discloses a spin-on glass (spino composition) using a perhydropolyazane having a weight average molecular weight of 4000 to 8000, a weight average molecular weight and a number average molecular weight ratio of 201132716 of 3.0 to 4.0. Patent Document 2 discloses a polyxazane-containing spin-coated glass having an average weight of 3000 to 6000. Further, the literature 3 discloses that the molecular weight in terms of polystyrene is 700 or less and the amount of sulfane is the total amount of polyazane. 1% or less of the coating liquid for cerium oxide-based coating. These are all intended to control the molecular weight of the polyazane, and the coating property is good, etc. According to the present inventors, if the weight average molecular total hydrogen polymerization is used, Although the arsenic has a tendency to be improved in embedding property, streaks are generated during coating, and when a perhydroalkane having a large weight average molecular weight is used, the generation of streaks is suppressed and the coating property is improved, but the embedding property tends to be small. When it is sufficiently buried in the deep part of the narrow gap, it is easy to cause the problem that the rate of hydrofluoric acid in the deep portion becomes large when the cerium oxide film is formed by firing, and there is such a problem. The problem is only the control time of the molecular weight distribution described in the patent documents 1 to 3, and further improvement is desired. [Prior Art Document] [Patent Document] [Patent Document 1] JP-A-2001-319927 [Patent Document 3] Japanese Patent Laid-Open No. Hei. No. Hei. No. Hei. No. Hei. No. Hei. No. Hei. 8 No. 7 specification n glass) Molecular weight, which is formed by a special polythene film, and the amount of change is small, and the etch point after the coating is degraded by coating is not charged 201132716 [Summary of the Invention] [Problems to be solved by the invention] As described above, in the conventional coating composition, when it is required to form a cerium oxide film for a substrate having a narrow gap, the embedding property and the coating property cannot be sufficiently achieved. The object of the present invention is to provide a coating which can sufficiently embed a narrow gap, in other words, a gap having a large aspect ratio, and which does not generate streaks during coating, for forming a cerium oxide film of a semiconductor device. Things. [Means for Solving the Problem] The coated composition of the present invention is a coated composition comprising perhydropolyazane and a solvent, characterized in that the molecular weight distribution curve of the perhydropolyazane is The range of the molecular weight of 800 to 2,500 and the molecular weight of 3,000 to 8,000 each have a maximum enthalpy, and the ratio Mw/Mn of the weight average molecular weight Mw to the number average molecular weight Μη is 6 to 12 . Further, a method for forming a ruthenium oxide film according to the present invention is characterized in that the method comprises the step of applying a coating composition on a surface of a substrate having irregularities, wherein the coating composition is a perhydropolyazoxide containing a coating composition formed with a solvent, wherein the molecular weight distribution curve of the perhydropolyazane has a maximum enthalpy in a range of a molecular weight of 800 to 2,500 and a molecular weight of 3,000 to 8,000, and a weight average molecular weight Mw and a number average molecular weight. The ratio Μη Mw/Mn is 6 to 12; and the coated substrate is subjected to heat treatment in an oxygen atmosphere of less than 1 〇〇〇 ° C or an oxidizing gas atmosphere containing water vapor to convert the composition into a hardening of the cerium oxide film. step. [Effects of the Invention] According to the coating composition of the present invention, the coating property and embedding property of the coating composition containing the polyazarane compound can be achieved, and the film physical properties of the obtained cerium oxide film can be improved. [Embodiment] [Embodiment of the Invention] Hereinafter, an embodiment of the present invention will be described in detail. Coating composition The coating composition according to the present invention is a solvent containing perhydropolyazane and a solvent capable of dissolving the perhydropolyazane. The perhydropolyazane used in the present invention is generally required to have a specific molecular weight and molecular weight distribution as described later, but the structure thereof is not particularly limited, and any structure can be selected without impairing the effects of the present invention. The perhydropolyazane which is an inorganic compound is composed of only ruthenium, nitrogen, and hydrogen, and is characterized in that impurities are less likely to be mixed when the ruthenium oxide film is formed by firing. The specific structure of such a perhydropolyazane can be represented by the following general formula (I). -(SiH2-NH) „- (I) In the formula, η represents the degree of polymerization. Further, a part or all of hydrogen containing a small amount of (I) may be substituted in a range not impairing the effects of the present invention. a polyazarane compound such as an alkyl group, an alkenyl group, a cycloalkyl group, an aryl group, an alkyl fluorenyl group, an alkylamino group or an alkoxy group. The polyazarane composition of the present invention contains the aforementioned It is a perhydropolyazane, but its molecular weight distribution curve of perhydropolyazane is extremely high in the region of molecular weight 800 to 2,500 and in the region of molecular weight 3,000 to 8,000. There is a minimum of 201132716 or more, preferably a very small enthalpy between the two maximes. The perhydropolyazane having such a molecular weight distribution curve can be modulated by any method, but the simplest can be mixed by A perhydropolyazane having a relatively large molecular weight, and a perhydropolyfluorene having a relatively small molecular weight, more specifically, preferably having a weight average molecular weight of 800 to 2,500, especially 1,000 to 2,200 all-hydrogen sand sands (hereinafter, simplified to low molecular weight It is obtained by mixing polypyrrolidine with a weight average molecular weight of 3,000 to 8,000, especially 3,500 to 7,000 (hereinafter referred to as "simplified by high molecular weight polyazane"). The method for synthesizing the perhydropolyazane before mixing is not particularly limited, but it can be synthesized by, for example, the method described in Patent Document 4 or 5. Not only the perhydropolyazane but also the polymer compound has a range in molecular weight distribution. When two polymer compounds having different molecular weights are mixed, the position of the maximum molecular weight distribution may change before and after mixing. This is particularly likely to occur when the molecular weight distributions of the two polymer compounds are similar. By mixing, the molecular weight with great enthalpy has a close tendency. It may become one depending on the situation. However, when mixing two perhydropolynitrazole plants having the weight average molecular weight as described above, the molecular weight difference In general, in general, the lanthanide does not become one. In addition, in the present invention, it is considered that by having the molecular weight between the two maximes The composition is reduced to exhibit the effects of the present invention, so it is necessary to mix at least two perhydropolyazane to have a very small enthalpy between the two maximes. In the present invention, two kinds of perhydropolythene are mixed in order to achieve the desired molecular weight distribution. In the case of azane, the molecular weight distribution of each perhydropolyazane is preferably 201132716. This is because the molecular weight distribution of the perhydropolyazane of either or both of the mixtures is wide. It is not easy to show a very small flaw between the two maximes of the curve, and the effect of the present invention is also reduced. Specifically, the two kinds of total hydrogen accumulation amount Mw and the number average molecular weight Μη before mixing are The high molecular weight component contained in the diazane alkane of the perhydropolyazane is simple, and a simple method of removing the high component is mentioned. That is, the lower the degree of solution of the perhydropolyazane, the smaller the molecular weight, the lower the solubility, and the solubility component of the perhydropolyazane alkane, which can be regarded as an indefinite component. The component which is dissolved in the solvent and separated by the insoluble component which is separated by filtration is a low molecular weight. Here, the perhydropolyazane is removed by a solvent to remove a high molecular weight, and the low molecular weight is removed. In such a method, although there are often or low molecular weight components, the mixture of compounds having a polymerization degree is narrow, which is a simple and effective method. The ratio of the weight average molecular weight Mw/Mn of each of Β 矽 alkane is preferably from 1.1 to 1.8. The distribution of the sub-quantity is narrow, and the molecular weight component and/or the low molecular weight are determined by the molecular weight dependence of the solubility by removing the perhydropolymerization and/or the low molecular weight component. Generally, the higher the molecular weight, the higher the solubility tends to be. The polymer is dissolved in a solvent having a solubility in the perhydropolypolymer, and the polymer-soluble low molecular weight component is separated by filtration by separating the insoluble component by filtration. That is, if the high molecular weight component is removed by removal, the amount of the component is removed. The solubility varies depending on the solvent to be used. After the components, the solubility may be further narrowed so that the width of the molecular weight distribution may not completely remove the high molecular weight component. In order to narrow the molecular weight distribution of different molecular weights, that is, different polymer compounds. For the solvent used for such a purpose, 201132716 is suitable, for example, as a hydrocarbon. For example, in the case of a hospital, as the carbon number increases, there is a tendency to dissolve the perhydropolyazane having a larger molecular weight. In general, hydrocarbons having a carbon number of about 5 to 10 can be used. Further, in order to narrow the molecular weight distribution of the polymer compound, it is also possible to separate the molecular weight of the perhydropolyazane by a conventionally used chromatography method or the like. However, if chromatography is used, the treatment time may become long, and from the viewpoint of productivity, it is preferred to use a method which is inferior in solubility to the aforementioned solvent. Further, it is also effective to carry out a treatment for narrowing the molecular weight distribution of perhydropolyazane, and to synthesize a perhydropolyazane having a narrow molecular weight distribution by performing a synthesis method or adjustment of a synthetic raw material. Further, in order to remove the high molecular weight component or the low molecular weight component contained in each perhydropolyazane before mixing the two perhydropolyazane, the molecular weight of the low molecular weight perhydropolysulfide is preferred. The component, in turn, is removed by the low molecular weight component of the high molecular weight perhydropolyazane. The effect of the present invention is more strongly exhibited by thus reducing the components of the intermediate regions of the two maximal enthalpies corresponding to the molecular weight distribution curve. The perhydropolyazane having different molecular weights is prepared in the above manner, and when mixed, the mixing ratio of the low molecular weight polyazane to the high molecular weight polyazane is preferably from 3:7 to 6:4. More preferably 4:6 to 6:4. If the mixing ratio is outside this range, the balance between coatability and embedding may be deteriorated. The perhydropolyazane having a specific molecular weight distribution in the present invention is generally as described above, and is obtained by mixing two kinds of perhydropolyazane-10-2011-032716 having different molecular weights, but it is also convenient. Obtained by other methods. For example, a desired molecular weight distribution can be achieved by preparing a full gas polysilicon plant having a broad molecular weight distribution and then removing the intermediate component having a molecular weight of around 2,500 to 3,000 by chromatography. Further, the perhydropolyazane system used in the present invention requires a ratio Mw/Mn of the weight average molecular weight Mw to the number average molecular weight Μη of 6 to 12, preferably 7 to 10%. The ratio of Mw/Mn is as described above. In the case of mixing two perhydropolyazane, the low molecular weight perhydropolyazane and the high molecular weight perhydropolyazane are 3:7. A weight ratio of 6:4 can be achieved when mixing. The coating composition of the present invention comprises a solvent which can dissolve the perhydropolyazane described above. The solvent is not particularly limited as long as it can dissolve the above-mentioned respective components. However, specific examples of preferred solvents include the following: (a) aromatic compounds such as benzene, toluene, and Toluene, ethylbenzene, diethylbenzene, trimethylbenzene, triethylbenzene, etc., (b) saturated hydrocarbon compounds such as n-pentane, isopentane, n-hexane, isohexane, n-heptane, isoheptane, n-octane , isooctane, n-decane, isodecane, n-decane, isodecane, etc., (c) alicyclic hydrocarbon compounds such as ethylcyclohexane, methylcyclohexane, cyclohexane, and rings Hexene, p-menthane, decalin, dipentene, limonene, etc., (d) ethers such as dipropyl ether, dibutyl ether, diethyl ether, methyl tertiary butyl ether, anisole, etc. (e) a ketone such as methyl isobutyl ketone or the like. Among these, (b) a saturated hydrocarbon compound, (c) an alicyclic hydrocarbon compound (d) ether, and (an anthrone) are used to adjust the evaporation rate of the solvent to reduce the pair. The harmfulness of human body -11 - 201132716 or the solubility of each component may be used, and two or more types may be used as appropriate. The coated composition used in the present invention may further contain other additives as needed. The component may, for example, be a viscosity modifier, a crosslinking accelerator, etc. Further, for the purpose of a semiconductor device, a phosphorusing effect or the like may be contained, and a phosphorus compound may be contained, for example, Further, the content of each component described above varies depending on the use of the intended composition, but the content of perhydropolyazane is preferably 10 to 25% by weight. It is preferably 12 to 22% by weight. In general, if the content of the perhydropolyazane is too high, the viscosity of the coating composition becomes high, the embedding property or the coating property tends to deteriorate, and if it is too low, it is formed. The thickness of the cerium oxide film is insufficient The method for producing a cerium oxide film according to the present invention can be formed on a substrate having a slit such as a groove or a hole, and can be sufficiently buried in the deep portion of the slit, the film surface is flat, and the film quality is uniform. Therefore, the transistor portion or the capacitor portion of the electronic device is formed as a planarization insulating film (pre-metal insulating film), and the yttrium oxide film is formed on the grooved substrate. The trench isolation structure may also be formed, and the trench isolation structure may be formed. The following describes the invention based on the method of forming the trench isolation structure. (A) Coating step The coating composition of the present invention is suitable for forming a substrate. For the groove isolation structure, when the groove isolation structure is formed, a substrate having a desired groove pattern of -12-201132716 is prepared. To form the groove, any method can be used, for example, as shown below. First, the film is formed on the surface of the ruthenium substrate by, for example, thermal oxidation. Here, the thickness of the ruthenium dioxide film formed is generally 5, as needed, on the formed ruthenium dioxide film. For example, the tantalum nitride film is formed. The tantalum nitride film can be used as a mask for subsequent etching or a termination layer in a honing step to be described later. Generally, it is formed by a thickness of 1 〇〇 to the thickness of the ruthenium dioxide film or the ruthenium nitride film formed by the film. If necessary, the photoresist film is dried or hardened, and then the desired light and development are used. Pattern formation. The exposure method can be carried out by any method such as exposure to a photomask. In addition, the photoresist can be selected from any viewpoint of resolution. The formed photoresist film is used as a mask, in order. The ruthenium dioxide film under the nitridation is etched, thereby forming a desired pattern on the tantalum nitride film and the second. The patterned tantalum nitride film and the ruthenium dioxide film are used as etched ruthenium substrates to form trench isolation trenches. form. The width of the trench isolation trench formed is determined by exposure of the photoresist film. The groove isolation groove in the semiconductor element differs depending on the purpose, but the width is generally 0.02 to 10 μηι, preferably μιη, and the depth is 200 to 1000 nm, preferably 300 to 700. In the past, a method for forming a trench isolation structure, but using a top-coating light film functioning at 400 nm by a cerium oxide to 30 nm CVD method for exposure, scanning, etc., a ruthenium film and a ruthenium oxide mask , dry light pattern conductor element 〇. 〇 5 to 5 nm. Compared with the present invention, it can be uniformly buried to a narrower and deeper portion from -13 to 201132716, so it is suitable for forming a narrower and deeper trench isolation structure. In particular, in the conventional method for forming a cerium oxide film-forming composition or a cerium oxide film, it is difficult to form a uniform cerium oxide film to a deep portion of the groove, and the width of the groove is generally 0.5 μm or less, especially 〇. In the case of forming a trench isolation structure having an aspect ratio of 5 or more, it is possible to uniformly form a ruthenium oxide film in the trench by using the composition for forming a ruthenium oxide film according to the present invention. Next, on the substrate thus prepared, a coating film of the coating composition formed of the material of the cerium oxide film was formed. The coating composition can be applied to the substrate by any method. Specifically, spin coating, curtain coating, dip coating, and the like are mentioned. Among these, spin coating is particularly preferred from the viewpoint of uniformity of the coating film surface and the like. In order to achieve both the groove embedding property and the coating property after application of the composition for forming a cerium oxide film, the thickness of the applied coating film is generally from 10 to 1,000 nm, preferably from 50 to 800 nm. The conditions of the coating vary depending on the concentration of the composition, the solvent, the coating method, and the like. However, the spin coating is exemplified as follows. In order to improve the production yield, a device is often formed on a large substrate. However, in order to uniformly form a coating film of a composition for forming a cerium oxide film on a ruthenium substrate of 8 Å or more, a plurality of stages of spin coating are combined. Is valid. First, a composition of 0.5 to 20 cc is dropped from each of the ruthenium substrates at a central portion of the ruthenium substrate or the entire surface of the substrate in such a manner as to form a coating film on average. -14- 201132716 Next, in order to extend the dripped composition to the entire surface of the substrate, it is rotated (pre-rotated) at a lower speed and for a short time, for example, a rotation speed of 50 to 500 rpm, 0.5 to 10 seconds. . Next, in order to make the coating film a desired thickness, it is rotated (main rotation) at a relatively high speed, for example, a rotation speed of 500 to 4500 rpm for 0.5 to 800 seconds. Further, in order to reduce the increase in the coating film at the peripheral portion of the substrate, and to dry the solvent in the coating film as much as possible, the rotation speed is faster than the rotation speed of the main rotation by more than 500 rPin, for example, the rotation speed. It is rotated (final rotation) at 1000 to 5000 rpm for 5 to 300 seconds. These coating conditions are appropriately adjusted depending on the size of the substrate to be used, the performance of the intended semiconductor element, and the like. (B) Hardening step After the coated composition is applied, a prebaking step may be carried out as needed. In the prebaking step, it is intended to completely remove the solvent contained in the coating film and to harden the coating film. In particular, in the method of forming a ruthenium oxide film of the present invention using a polyazide-containing composition, the pre-baking treatment is performed to enhance the density of the formed cerium oxide film, so that it is preferable to combine pre-baking. step. Generally, the pre-baking step is a method in which heating at a fixed temperature is substantially employed. Further, in the curing, in order to prevent the coating film from shrinking, the slit portion to become a dent, and a void in the slit, it is preferable to control the temperature in the prebaking step and raise it with time while prebaking. The temperature in the pre-baking step is usually in the range of 50 ° C to 400 ° C, preferably in the range of 1 〇〇 to -15 - 201132716 3 0 (TC). The time required for the pre-baking step is generally 10 seconds. 30 minutes, preferably 30 seconds to 10 minutes. To increase the temperature in the prebaking step over time, a method of gradually increasing the temperature of the gas atmosphere in which the substrate is placed, or monotonously increasing the temperature Here, the highest prebaking temperature in the prebaking step is generally set to be higher than the boiling point of the solvent for the cerium oxide film forming composition from the viewpoint of removing the solvent from the coating film. In addition, in the method of the present invention, before the pre-baking step is combined, the substrate which becomes higher temperature than the pre-baking is lowered, preferably 5 〇 ° C or higher, and the highest during pre-baking. The substrate having a temperature lower than the temperature is preferably subjected to a hardening step. By applying the hardening step to the substrate before the temperature is lowered, the energy and time for further increasing the temperature can be saved. Next, in order to convert the polypyrazine-containing coating film into oxidation Membranous membrane The curing is performed by heating the entire substrate and applying a curing step. Generally, the entire substrate is usually placed in a curing furnace or the like and heated. The curing is preferably a hardening furnace or a heating plate, and an inert gas or oxygen gas containing water vapor. The water vapor is important in sufficiently converting the polyazane to a cerium oxide film (i.e., cerium oxide), preferably 30% or more, more preferably 50% or more, and most preferably When the water vapor concentration is 80% or more, the conversion of the organic compound to the cerium oxide film is facilitated, and defects such as voids are less likely to be formed, and the characteristics of the cerium oxide film are improved. As the gaseous ambient gas, nitrogen, argon, helium, etc. are used. The temperature conditions for curing are changed depending on the type of the composition or the combination of the steps. However, in the higher case, the rate at which polyazane is converted into a ruthenium oxide film is faster. In addition, when the temperature is low, the oxidation or crystal structure of the ruthenium substrate has a small adverse effect on the device characteristics. Therefore, in the present invention, the hardening step in the present invention is usually carried out at a temperature of from 400 to 700 ° C below 100 ° C. Here, the rise to the target temperature is generally from 1 to 100 ° C /min. The hardening after reaching the target temperature is 1 minute to 1 hour, preferably 15 minutes to 3 hours. The composition of the hardening temperature or the hardened gas environment can be changed stepwise by heating, and the polydecazane is converted. Formation of cerium oxide to form cerium oxide The method for forming a cerium oxide film according to the present invention is necessary for each step, but a further step such as honing step or etching may be combined as needed. EXAMPLES The present invention is as follows: Synthesis of Example 1 Synthesis of Low Molecular Weight Polyazane A 400 g of anhydrous pyridine having a purity of 99% or more of dichlorosilane was stirred while being 5 kg of 0 °C. While maintaining the temperature of the mixture at 0 ° C while stirring 1.22 kg of ammonia gas having a purity of 99.9%, the mixture was mixed until the temperature of the mixture was maintained at 〇 ° C while stirring was continued: hourly reaction. Dry nitrogen was blown into the reaction mixture for 30 minutes, and the remaining ammonia was then filtered and separated from the slurry-like reaction mixture to the filtrate A. The xylene was mixed into the obtained filtrate A and heated to remove pyridine under a reduced pressure of 20 mmHg to prepare a 20% by weight solution of a polymer having a molecular weight of 1450. The temperature tends to be the result, and the preferred temperature time is as needed. By the plasma membrane. The aforementioned inscription steps are injected into the contemporaneous material. After removing 1% of ammonium, the obtained ammonium salt was obtained at 50 ° C, and the amount was averaged -17-201132716. The obtained 20% by weight xylene solution was heated to 50 ° C, and xylene was distilled off under a reduced pressure of 10 mmHg. The positive solution was poured into the obtained colorless transparent liquid to prepare a white solution having a concentration of 1% by weight. This solution was filtered with a filter having a filtration precision of 〇 2 μm to obtain a polymer solution. Dibutyl ether was mixed with the polymer solution and heated to 50 ° C, and n-pentane was distilled off under a reduced pressure of 2 OmmHg to prepare a ratio MW containing a weight average molecular weight of 1,100, a weight average molecular weight Mw and a number average molecular weight Μη. /Mn is a polymer solution 1 of a concentration of 20% by weight of the polymer of 1.45. Synthesis Example 2 Synthesis of high molecular weight polysallium Hydrazine The filtrate A was prepared in the same manner as in Synthesis Example 1, and further heated in a sealed system at 150 ° C for 3 hours. After cooling to room temperature, it was returned to normal pressure, xylene was mixed to the obtained solution and heated to 50 ° C, and pyridine was distilled off under a reduced pressure of 20 mmHg to prepare a polymer having a weight average molecular weight of 6000. A solution of weight % concentration. The obtained 20% by weight xylene solution was heated to 50 ° C, and xylene was distilled off under reduced pressure of 10 mmHg. n-Heptane was added to the obtained white powder to prepare a dispersion of 1% by weight. The dispersion was filtered under reduced pressure using a glass filter (manufactured by ADVANTEC Toyo Co., Ltd.: GF-75 (trade name)), and the solvent was removed. The obtained white powder was dissolved in dibutyl ether to prepare a polymer solution having a concentration of 20% by weight of a polymer having a weight average molecular weight of 64 Å, a weight average molecular weight Mw and a number average molecular weight Μη Mw/Mn of 1.22. . Synthesis Example 3 Synthesis of Ultrahigh Molecular Weight Polyazane -18 - 201132716 The filtrate A was prepared in the same manner as in Synthesis Example 1, and heated in a sealed system at 150 ° C for 6 hours. After cooling to room temperature, returning to normal pressure, dibutyl ether was mixed to the obtained solution and heated to 5 (TC, distillation was carried out to remove pyridine under a reduced pressure of 20 mmHg, and a polymer having a weight average molecular weight of 9,200 was prepared. 2% by weight concentration of the polymer solution 3. Example 1 Mix 60 g of the polymer solution 1 and 40 g of the polymer solution 2. The mixed polymer solution has a molecular weight distribution curve of 6300 and 650 It has a maximum enthalpy and Mw/Mn is 10. A substrate having a groove having a width of 0.5 μm and having a width of 0.05, 0.1, 0.2, and 0.5 μm and having a surface covered with a tantalum nitride inner liner is prepared as a slab. The prepared polymer solution was applied as a coating composition by spin coating on the substrate. The coating conditions were set to pre-rotation: 300 rpm/5 seconds, main rotation: 1 000 rpm / 20 seconds, final rotation: 1500 rpm /10 seconds. After observing the film surface after coating, it was confirmed that no streaks were generated, and excellent coating properties were obtained. Further, the coated substrate was prebaked on a hot plate at 150 ° C for 3 minutes, and continued to be cooled. Import The firing furnace in a pure oxygen atmosphere is fired in a firing furnace at a heating rate of 10 ° C/min to 800 ° C, and further fired in an oxygen gas atmosphere having a water vapor concentration of 80% for 30 minutes. After the FT-IR of the film was formed, the absorption of the wave number 1080 cnT 1 attributed to the Si-Ο bond was observed, and it was confirmed that the ruthenium oxide film was obtained. On the other hand, the wave attributed to the NH bond and the Si-H bond was not observed. The absorption of 3380 cm·1 and 2200 cm·1 confirmed that the perhydropolyazane had been converted into cerium oxide. -19- 201132716 In addition, 0.5% by weight of hydrofluoric acid and 30% by weight of ammonium fluoride were used. The aqueous solution was etched at 23 ° C as an etching solution, and the relative etching rate of the thermally oxidized yttrium oxide film was measured to be 1.48. After the fired substrate was cut at a right angle to the longitudinal direction of the groove, It was immersed in an aqueous solution containing 0.5% by weight of hydrofluoric acid and 30% by weight of ammonium fluoride for 30 seconds, and further washed thoroughly with pure water and then dried. The cross section of the substrate was scanned by a scanning electron microscope. 50,000 times and viewed from an elevation angle of 30° from the direction perpendicular to the cross section The etching amount was evaluated at the deepest portion of the groove. Even if the groove width was changed, the change in the uranium amount was only a little, and it was confirmed that a dense ruthenium oxide film was sufficiently formed even in the deepest portion of the groove of 0.05 μη wide. Example 2 40 g of the polymer solution 1 and 60 g of the polymer solution 2 were mixed. The polymer solution after mixing was at a molecular weight distribution curve having a molecular weight of 6250 and a position of 680 having a maximum enthalpy and a Mw/Mn of 10 The prepared polymer solution was used as a coating composition, and was applied in the same manner as in Example 1 and applied onto a ruthenium substrate. After observing the film surface after coating, it was confirmed that no streaks were generated, and excellent coatability was achieved. Further, this coated substrate was fired in the same manner as in Example 1. After the FT-IR of the obtained fired film was measured, the absorption of the wave number of 1 080 cnT1 attributed to the Si-Ο bond was observed, and it was confirmed that the ruthenium oxide film was obtained. On the other hand, the absorption of the wave number 3 3 8 0 cm·1 and 2 2 00 cnT1 attributed to the N-H bond and the Si-H bond was not observed, and it was confirmed that the perhydropolyazane was converted into cerium oxide. In the same manner as in the first embodiment, the phase etch rate of the oxidized cerium oxide film was measured to be 1.50. The cross section of the fired substrate was observed in the same manner as in Example 1, and the etching amount was evaluated. Even if the groove width was changed, the change in the etching amount was only a little, and it was confirmed that a dense ruthenium oxide film was sufficiently formed even in the deepest portion of the groove of 0.05 μη wide. Comparative Example 1 Polymer solution 1 was used as a coating composition, and was applied in the same manner as in Example 1 and applied onto a ruthenium substrate. After observing the film surface after coating, it was confirmed that a plurality of stripes were generated from the center portion toward the peripheral portion, and the coatability was insufficient. Further, this coated substrate was fired in the same manner as in Example 1 and the cross section was observed with an electron microscope. In the groove of 1 μm or more, the amount of etching was changed only a little, and it was confirmed that a dense ruthenium oxide film was sufficiently formed even in the deepest portion. However, in the deepest portion of the groove of 〇5〇ηι, the amount of etching was large, and it was confirmed that a dense ruthenium oxide film was not formed in this portion. Comparative Example 2 The polymer solution 2 was used as a coating composition, and was applied in the same manner as in the Example, and applied onto a ruthenium substrate. After observing the film surface after coating, it was confirmed that no streaks were generated, and excellent coatability was achieved. Further, the coated substrate was fired in the same manner as in Example 1, and the cross section was observed with an electron microscope. In the groove having a width of more than 0.2 μm, the change in the etching amount was only a little, and it was confirmed that a dense cerium oxide film was sufficiently formed even in the deepest portion. However, in the deepest portion of the groove of 0.05 μηιη wide and 〇.] -21-201132716 μπι, the amount of etching was large, and it was confirmed that the portion did not form a dense cerium oxide film. Comparative Example 3 The polymer solution 3 was used as a coating composition, and was applied in the same manner as in Example 1 and applied onto a ruthenium substrate. After observing the film surface after coating, it was confirmed that no streaks were generated, and excellent coatability was achieved. Further, this coated substrate was fired in the same manner as in Example 1 and the cross section was observed with an electron microscope. In the groove having a width of 0.2 μm or less, the void is recognized, and the space for improvement in embedding property can be confirmed. [Simplified description of the drawing] ^ΙΊΓ None. [Main component symbol description] 4nt pick. •twenty two-