JP4623809B2 - Wafer surface inspection apparatus and wafer surface inspection method - Google Patents

Description

【００５０】
この式は各角度の差の内積を求めるものであり、角度類似度Ｚが１に近づくほど２本のブロック３２ａ及び３２ｂの連結度、すなわち接続すべき度合いが高くなる。
【００５１】
これは２本のブロック３２ａ及び３２ｂの角度θ１及びθ２が近似しており、しかも２本のブロック３２ａ及び３２ｂが直角方向に離れていないこと（角度θ３が小さいこと）を条件としてこれら２本のブロック３２ａ及び３２ｂを連結すべきであると判断することを意味している。但し、この式において連結度が高いと判断された場合であっても、２本のブロック３２ａ及び３２ｂの間隔が大きい場合にはこれらを連結すべきではなく、判定用コンピュータ２１は図９に示す方法によって２本のブロック３２ａ及び３２ｂを連結すべきであるか否かを判断する。
【００５２】
すなわち図９において、判定用コンピュータ２１は２本のブロック３２ａ及び３２ｂの共通の近似直線Ｌ２を引き、２本のブロック３２ａ及び３２ｂの近似直線Ｌ２の方向の間隔Ｌ３が予め設定されている所定値よりも小さい場合、２本のブロック３２ａ及び３２ｂを連結すべきものと判断する。
【００５３】
このように２本のブロック３２ａ及び３２ｂの近似直線Ｌ２の方向の間隔Ｌ３に基づいて連結の必要性を判断することにより、ブロック３２ａ及び３２ｂの最も近い２点間の距離Ｌ４に基づいて判断する場合に比べて、連結すべき近接した２本のブロック３２ａ及び３２ｂを確実に連結することができる。
【００５４】
次に図４に示したシリコンウエハ表面の散乱点３１の偏在（ヨゴレ）の検出処理ステップＡ１３における詳細な処理手順を説明する。図１０はシリコンウエハの表面の散乱点３１の検出状態を示すＬＰＤマップ３０を示し、散乱点３１の偏在領域３５ａ、３６ａ、３７ａが存在している状態を示す。この状態においては、ＬＰＤマップ３０の散乱点３１は図１１（Ａ）に示すようなドットとして表されている。このようなＬＰＤマップ３０の各ドットに対して、判定用コンピュータ２１は例えば２５６階調のビットマップへの変換処理を施すことにより、図１１（Ｂ）に示すようなビットマップデータＢＭを得る。
【００５５】
判定用コンピュータ２１はビットマップデータＢＭに対して空間フィルタを用いた平滑化処理を施すことにより、図１１（Ｃ）に示すようにビットマップデータＢＭを平滑化してなる平滑化曲線Ｓ１を得る。かかる平滑化曲線Ｓ１を画像として表すと、図１０（Ｂ）に示すように、散乱点３１の偏在領域３５ａ、３６ａ、３７ａのみがその周囲がぼやけた状態で表される。この表示状態では、平滑化処理の結果として、各ドットのうち他のドットから離れたものほど薄く表されることとなる。
【００５６】
そして判定用コンピュータ２１は図１１（Ｃ）に示すように、かかる平滑化曲線Ｓ１を予め設定されている閾値ＳＨで２値化することにより、２値化領域データＤ３５Ｃを得る。この２値化領域データＤ３５Ｃによって図１０（Ｃ）に示すような周囲とは明確に輝度が異なる偏在領域３５ｃ、３６ｃ、３７ｃを得る。このように散乱点３１の偏在の領域について閾値ＳＨを設けて抽出することにより、周囲に比べて僅かに散乱点の密集度が高い領域であっても、閾値ＳＨの選択によって当該偏在領域を確実に抽出することができる。
【００５７】
またこのように２値化データＤ３５Ｃを得る際の閾値ＳＨを適宜変更することにより、シリコンウエハ表面の散乱点３１の検出状態（散乱点３１の全体的な密度）に応じた偏在領域の検出を行うことができる。例えば散乱点３１が全体的に多く検出された場合には閾値ＳＨの設定レベルを高くすることにより、特に検出密度の高い領域を偏在領域として、散乱点３１が平均的に存在する他の領域と区別して検出することが可能となる。そして判定用コンピュータ２１はこのようにして検出された不定形の偏在領域をヨゴレとして認識するとともに、これを画像情報や位置、形状等を表す数値情報（ウエハの表面情報）として保持する。
【００５８】
以上のようにシリコンウエハ表面のキズやヨゴレが検出されると、判定用コンピュータ２１はこれらの検出されたキズやヨゴレが不良とすべきものであるか否かを、図４に示した処理ステップＡ１４において判断する。
【００５９】
すなわち、判定用コンピュータ２１は図４のステップＡ１２において抽出された線状のキズについて、その長さと検出強度とに基づいて不良であるか否かを判定するようになされている。この場合、キズの長さは図８及び図９について上述した方法により１本と認識されたキズの長さを意味し、また、キズの検出強度とは図６及び図７について上述した回転プロジェクション法におけるプロジェクション曲線ＸＰ、ＹＰ（ヒストグラム）のピーク値ＰＥＡＫ（キズの濃さに相当）を意味する。
【００６０】
また判定用コンピュータ２１は図４のステップＡ１３において抽出された偏在（ヨゴレ）について、その外接四角形を求め、当該外接四角形の面積、縦方向の長さ、横方法の長さ、対角線の長さ、密度及び面積-密度判定曲線に基づいて不良であるか否かを判定するようになされている。すなわち図１２に示すように、上述のステップＡ１３において抽出された偏在（ヨゴレ）領域３５ｃについて、判定用コンピュータ２１はその外接四角形３５ｄを求め、当該外接四角形３５ｄの縦方向の長さＨ、横方向の長さＷ及び対角線の長さＤ及び面積を測り、さらに当該偏在（ヨゴレ）領域３５ｃにおける各散乱点３１の積分値に基づいて密度を求める。
【００６１】
そして判定用コンピュータ２１は外接四角形３５ｄの面積及び密度に基づいて、（密度−漸近密度）＞係数α／（面積−漸近面積）によって表される面積-密度判定式を満たす場合に不良と判定する。これは図１３に示すような面積-密度判定曲線Ｓ３５よりも密度及びまたは面積の値が大きい場合に不良と判定することを意味している。すなわち、人がヨゴレていると判定する条件として、面積と密度とが反比例の関係にあることに着目し、このことを条件式として表したものが上述の面積-密度判定式となる。この条件式を用いることにより、従来、人が主観で判断していたヨゴレによる不良の判定を同一条件で行うことが可能となる。
【００６２】
判定用コンピュータ２１は係るキズ及びヨゴレに基づく不良の基本的判定方法に加えて、キズ及びヨゴレの種類をその散乱点の集合領域、偏在領域の特徴量に応じて分類し、当該分類結果に応じて不良判定の基準設定を行うようになされている。すなわち図１４に示すように、シリコンウエハの表面に形成されているキズには、例えばラッピング処理において発生する円弧状の規則正しい配列のキズ（以下これを第１のキズと呼ぶ）３９や、電気的耐圧性（ＧＯＰ）の劣化を伴うポックマークと呼ばれる欠陥の集合体からなる不規則な配列のキズ（以下これを第２のキズと呼ぶ）４０があり、第１のキズ３９よりも第２のキズ４０の方を重度の欠陥と判断する必要がある。
【００６３】
従って判定用コンピュータ２１は図４に示した判定処理ステップＡ１４において、キズの種類を散乱点の集合状態（特徴量）に応じて判断するとともに、当該キズの種類に応じて判定基準を変えるようになされている。例えば、特徴量である散乱点の配列が比較的不規則な第２のキズに対しては、特徴量である散乱点の配列が規則的な第１のキズの場合に比べてその判定基準とする長さを短くすることにより、第１のキズでは不良とはならない長さであっても第２のキズでは不良となるようになされている。
【００６４】
また、判定用コンピュータ２１は図４に示した判定処理ステップＡ１４において、ヨゴレの種類を図１２について上述した散乱点の集合状態（特徴量）に応じて判断するとともに、当該ヨゴレの種類に応じて図１３について上述した判定基準（面積-密度判定曲線Ｓ３５）を変えるようになされている。
【００６５】
因みに、図１５（Ａ）は判定用コンピュータ２１によるキズ及びヨゴレの検出方法を示した図であり、また図１５（Ｂ）は不良ウエハの判定基準を示した図である。ここではキズやヨゴレを種類分けするための特徴量としてその長さ、濃さ、面積、密度を用いたが、キズの特徴量としては、その長さ、密度、幅、直線度、円弧度、位置等があり、またヨゴレの特徴量としては、その面積、濃さ/密度、分布、形状、位置等があり、判定用コンピュータ２１は必要に応じてこれらの特徴量を使い分ける。
【００６６】
かくして判定用コンピュータ２１によってシリコンウエハ表面のキズやヨゴレに基づく不良の判定が行われ、その結果がプリンタ等において出力されることとなる。
【００６７】
判定用コンピュータ２１を有する表面検査システム１０は、かかるキズ及びヨゴレの検出（ウエハ表面情報の生成）並びに不良ウエハの判定処理に加えて、キズやヨゴレが発生した工程を特定する機能を有している。
【００６８】
すなわち、図１との対応部分に同一符号を付して示す図１６は、レーザ散乱検出機（ウエハ表面検査装置）１１（１１Ａ、１１Ｂ、１１Ｃ）を有する各抽出部１３（１３Ａ、１３Ｂ、１３Ｃ）をウエハ１００の製造工程の各工程（例えば表面研磨工程である第１工程及び洗浄工程である第２工程）の各前後に配置した場合の構成例を示すブロック図である。
【００６９】
この図１６において、各レーザ散乱検出機１１Ａ、１１Ｂ及び１１Ｃによって検出されたＬＰＤマップＬＰＤ１、ＬＰＤ２及びＬＰＤ３はそれぞれ判定用コンピュータ２１のハードディスクに各工程情報に対応付けられて記憶される。すなわち、各レーザ散乱検出機１１Ａ、１１Ｂ及び１１Ｃから各ウエハ毎に出力される各ＬＰＤマップＬＰＤ１、ＬＰＤ２及びＬＰＤ３には、当該ＬＰＤマップのウエハＩＤ、スロット番号等が付随しており、判定用コンピュータ２１はこれらウエハを特定する情報及びこれに加えてＬＰＤマップの出力元であるレーザ散乱検出機１１Ａ、１１Ｂまたは１１Ｃを特定する情報（レーザ散乱検出機１１Ａ、１１Ｂ及び１１Ｃの各ＩＤ等）をＬＰＤマップとともに記憶している。
【００７０】
判定用コンピュータ２１において記憶されたＬＰＤマップの出力元であるレーザ散乱検出機を特定する情報は、その情報に対応するウエハのウエハＩＤやスロット番号等に対応付けられて時系列に判定用コンピュータ２１に蓄積されて行く。したがって、判定用コンピュータ２１はあるウエハについての蓄積されたレーザ散乱検出機のＩＤ（これら蓄積された情報を工程履歴情報と呼ぶ）によって当該ウエハの工程履歴及び現在時点での最終工程を確認することができる。
【００７１】
従って、例えば第1工程の前に設けられたレーザ散乱検出機１１Ａから出力されたＬＰＤマップＬＰＤ１に対応する工程履歴情報は、なにも蓄積されていない状態であり、判定用コンピュータ２１はこのことからＬＰＤマップＬＰＤ１が第１工程の前におけるあるウエハのＬＰＤマップであることを認識することができ、また第２工程の前に設けられたレーザ散乱検出機１１Ｂから出力されたＬＰＤマップＬＰＤ２に対応する工程履歴情報は、第１のレーザ散乱検出機１１Ａを表す情報のみが蓄積されている状態であり、判定用コンピュータ２１はこのことからＬＰＤマップＬＰＤ２が第１工程終了時点、すなわち第２工程前におけるあるウエハのＬＰＤマップを示すものであることを認識することができる。
【００７２】
因みに、工程情報としては各レーザ散乱検出機１１のＩＤを蓄積して利用することに代えて、図１６に示すように、各工程において処理したウエハのウエハＩＤやスロット番号を工程を表す情報とともに工程情報Ｄ（ＰＲＯ）としてネットワーク１５を介して判定用コンピュータ２１に供給するようにしてもよい。この場合、判定用コンピュータ２１は各工程から供給されるウエハＩＤによって当該ウエハＩＤで特定されるウエハがどの工程を終了した状態であるか、及び当該ウエハの工程履歴を認識することができる。
【００７３】
かくして、各レーザ散乱検出機１１から供給された各ＬＰＤマップは、それぞれのウエハの工程履歴情報に対応付けられて判定用コンピュータ２１に蓄積される。当該蓄積されたＬＰＤマップは、図４について上述した処理工程によってキズやヨゴレの検出に利用される。
【００７４】
図４の処理手順において検出されたキズ及びヨゴレの情報（ウエハの表面情報）は、図１４や図１０（Ｃ）に示した画像情報やその位置、大きさ、濃さ等の数値情報として判定用コンピュータ２１において生成されており、当該判定用コンピュータ２１はこれらの表面情報を同一の工程履歴のウエハについて、重畳するようになされている。重畳された結果は支援装置１２１に供給され、キズやヨゴレが発生した工程を特定する際に利用される。
【００７５】
すなわち、図１７は判定用コンピュータ２１及び支援装置１２１による不良工程の特定処理手順を示し、判定用コンピュータ２１はステップＡ２１から当該処理手順に入ると、ステップＡ２２において、図４について上述したキズ及びヨゴレの検出処理で既に生成されているウエハの表面情報を、同一の工程情報のウエハのものについて重畳する。
【００７６】
この結果、図１８に示すように、同じ位置に生じているキズは同じ位置で重畳されて行くことにより、ある範囲（図１８において領域１４０、１４１、１３０、１３１）でのキズ情報、ヨゴレ情報の発生頻度が高くなると、このことはこのとき表面情報を重畳している各ウエハの工程履歴（このとき表面情報を重畳している各ウエハの工程履歴は同一である）のいずれかにおいてウエハを不良とするようなキズやヨゴレが恒常的に発生しているということになる。
【００７７】
従って判定用コンピュータ２１は、重畳結果をステップＡ２３においてモニタに画像として表示した後、この重畳データを支援装置１２１に供給し、支援装置１２１においてステップＡ２４の不良工程特定処理を実行する。
【００７８】
この特定処理において、支援装置１２１は、各工程履歴情報に対応した表面情報を重畳した各重畳データのうち、最終工程が１工程分異なる重畳データ（例えば図１９に示す重畳データＤｎ、Ｄｍ）を比較する。例えば、図１９に示すように、工程履歴が第１工程だけの表面情報Ｄ１、Ｄ２、……を重畳してなる重畳データＤｎ（すなわち最終工程が第１工程のもの）と、第１工程及び第２工程の表面情報Ｄ１１、Ｄ１２、……を重畳してなる重畳データＤｍ（すなわち最終工程が第２工程のもの）では、第２工程が含まれているか否かの差異がある。従って、支援装置１２１は、これらの重畳データを比較し、発生頻度の高いキズやヨゴレが一方の重畳データにはなく、他方の重畳データにある場合、第２工程にキズやヨゴレの発生原因があると判定することができる。
【００７９】
支援装置１２１は、かかる判定結果及び前記重畳結果をハードディスク等の記憶媒体に記憶させるとともに、これらをモニタやプリンタに出力した後、ステップＡ２５において当該処理手順を終了する。
【００８０】
因みに、判定に使用する重畳データは上述したように異なるウエハのものを用いることに代えて、同一のウエハの重畳データを工程履歴が加算されるごとに保存しておき、当該保存された同一ウエハについての重畳データを比較するようにしてもよい。
【００８１】
[動作]
上述のような機能、構成を有する本発明に係るシリコンウエハの表面情報処理装置（判定用コンピュータ２１及び支援装置１２１）においては、レーザ散乱検出機１１から供給されたＬＰＤマップに基づいてキズ及びヨゴレを画像情報や種々の数値等からなる情報として生成した後、当該生成された表面情報を複数のウエハについて重畳する。この場合、ウエハ表面のある決まった位置にキズやヨゴレが多く発生する傾向があると、重畳結果として当該位置において同じようなキズやヨゴレの重畳量が多くなる。
【００８２】
従って当該重畳結果を見ることにより、個々のウエハについて検出されたキズやヨゴレが、そのウエハのみに偶発的に発生したものであるか、または恒常的に発生するものであるかを判断することが可能となる。
【００８３】
キズやヨゴレが発生する工程は、各ウエハの表面情報に対応付けられた工程履歴情報によって特定が可能であり、支援装置１２１は当該工程履歴情報によって、キズやヨゴレが発生している工程を特定する。従ってオペレータは、当該特定結果を見るだけで欠陥発生工程を簡単に見つけることができる。そして、オペレータは当該見つけられた欠陥発生工程を検査改善することにより、不良品の連続的な発生をいち早く除去でき、作業工程を改善することが可能となる。
【００８４】
[他の実施形態]
なお上述の実施形態においては、図１７について上述したように、判定用コンピュータ２１においてＬＰＤマップに基づいて検出されたキズ及びヨゴレを画像情報や種々の数値情報として生成し、当該生成された情報（表面情報）を重畳して重畳データを得る場合について述べたが、本発明はこれに限らず、レーザ散乱検出機１１から供給されるＬＰＤマップをウエハの表面情報として判定用コンピュータ２１及び支援装置１２１によって重畳するようにしてもよい。この場合の処理手順を図２０に示す。
【００８５】
すなわち、図２０は判定用コンピュータ２１及び支援装置１２１による不良工程の特定処理手順の他の実施形態を示し、判定用コンピュータ２１はステップＡ３１から当該処理手順に入ると、ステップＡ３２において、図３について上述したようなレーザ散乱検出機１１から供給されたＬＰＤマップを、同一の工程情報のウエハのものについて重畳する。
【００８６】
この結果、図２１に示すように、同じ位置に生じているＬＰＤは同じ位置で重畳されて行くことにより、ある範囲（図２１において領域２４０、２４１、２３０、２３１）でのＬＰＤの発生頻度が高くなると、このことはこのときＬＰＤマップを重畳している各ウエハの工程履歴のいずれかにおいてウエハを不良とするようなキズやヨゴレが恒常的に発生していることを表していることになる。因みにこの場合の重畳されたＬＰＤに基づくキズ及びヨゴレの検出方法は、図１〜図１５について上述した方法を用いる。
【００８７】
判定用コンピュータ２１は、重畳結果及び検出結果をステップＡ３３においてモニタに画像として表示した後、この重畳データを支援装置１２１に供給し、支援装置１２１においてステップＡ３４の不良工程特定処理を実行する。この不良工程特定処理は図１７について上述した処理ステップＡ２４と同様である。
【００８８】
かくして、判定用コンピュータ２１及び支援装置１２１は、ＬＰＤマップを直接重畳することによってもキズやヨゴレの発生原因である工程を特定することができる。このようにＬＰＤマップを直接重畳した場合、キズやヨゴレといった外的要因による不良のほかに、個々のウエハではキズやヨゴレとは見なされないような微小なＬＰＤが同じ位置に恒常的に発生するといった欠陥（図２１に示す領域２４２）も検出することが可能となる。
【００８９】
また上述の実施形態においては、工程履歴情報としてウエハが経た工程を記録して行く場合について述べたが、本発明はこれに限らず、例えば現時点での最終工程のみを記録する等、種々の方法を適用することができる。
【００９０】
また上述の実施形態においては、判定用コンピュータ２１においてウエハの表面情報を重畳し、当該重畳してなる重畳データを支援装置１２１に供給して欠陥発生工程を特定する場合について述べたが、本発明はこれに限らず、判定用コンピュータ２１において検出及び生成されたキズ情報及びヨゴレ情報（ウエハの表面情報）を支援装置１２１に供給し、支援装置１２１において表面情報を重畳するようにしてもよい。
【００９１】
また上述の実施形態においては、ウエハの製造工程として表面研磨工程及び洗浄工程を例にして説明したが、本発明はこれに限らず、例えばウエハの搬送工程等も工程として本発明を適用することができる。このようにすれば、搬送時におけるキズやヨゴレの発生を判定することが可能となる。この場合、支援装置１２１は工程履歴情報及びこれに対応したウエハ表面情報を判定用コンピュータから取得し内部のハードディスク等の記憶媒体に蓄積するとともに、当該蓄積された情報に基づいてウエハ搬送時に掛ける保険金額を判断することが可能となる。
【００９２】
また、上述の実施形態においては、判定用コンピュータ２１及び支援装置１２１を１セット有するシステムについて述べたが、本発明はこれに限らず、複数の判定用コンピュータ２１及び支援装置１２１によってそれぞれの表面情報や重畳データを共有するようにしてもよい。
【００９３】
また、上述の実施形態においては、キズやヨゴレの検出及びこれらが発生した工程の特定を行うシステムについて述べたが、本発明はこれに限らず、例えば判定用コンピュータ２１または支援装置１２１において需要者のウエハに対する要求品質を保持しておき、その要求品質よりも品質が劣化していると判断されるウエハを判定し、このウエハに発生している欠陥が生じた工程を特定するようにしてもよい。この場合、品質とは図４について上述した判定処理におけるキズやヨゴレの形状、濃さ等を基に判断されるものである。
【００９４】
また上述の実施形態においては、複数の工程の間にそれぞれレーザ散乱検出機１１Ａ、１１Ｂ及び１１Ｃを設け、これら複数のレーザ散乱検出機１１Ａ、１１Ｂ及び１１Ｃから得られるＬＰＤマップ（すなわち各工程間において検出される複数のＬＰＤマップ）に基づいて不良が発生する工程を特定する場合について述べたが、本発明はこれに限らず、例えば全工程または所定工程単位のうちの最終工程後に設けられたレーザ散乱検出機１１Ｃから得られるＬＰＤマップのみによって不良発生工程を判断するようにしてもよい。
【００９５】
この場合、判定用コンピュータ２１または支援装置１２１において、各工程において発生する不良内容（キズやヨゴレの位置、曲線キズの半径等）の傾向を予め記憶しておき、最終工程の後に検出されるＬＰＤマップを図１７または図１９について上述した場合と同様にして重畳するとともに当該重畳結果に基づいてキズやヨゴレの情報（ウエハの表面情報）を検出する。そして、このようにして検出されたウエハの表面情報のうち、予め記憶している不良内容の傾向に合致するものに基づいてそのような不良発生傾向を有する工程を特定する。このようにすれば、各工程間のそれぞれにレーザ散乱検出機を設けることなく、不良発生工程を特定することが可能となる。因みに、各工程ごとの不良発生傾向を表す情報としては、キズやヨゴレの特徴量として上述した項目（キズの特徴量としては、その長さ、密度、幅、直線度、円弧度、位置等があり、またヨゴレの特徴量としては、その面積、濃さ/密度、分布、形状、位置等がある）を用いることができる。
【００９６】
【発明の効果】
以上説明したように、本発明に係るシリコンウエハの表面情報処理装置は、ウエハの表面情報を重畳することによって恒常的に発生しているキズやヨゴレを検出することが可能であり、さらにその発生工程を検出することも可能となる。従って検査効率の向上をもたらすことが可能となる。
【図面の簡単な説明】
【図１】 本発明に係るシリコンウエハの検査システムの全体構成を示す略線図である。
【図２】 本発明に係る欠陥の判定用コンピュータの構成を示すブロック図である。
【図３】 ＬＰＤマップの検出例を示す略線図である。
【図４】 本発明に係るシリコンウエハの表面検査処理手順を示すフローチャートである。
【図５】 本発明に係るキズの検出処理におけるピラミッド処理の説明に供する略線図である。
【図６】 本発明に係るキズの検出処理における回転プロジェクション法の説明に供する略線図である。
【図７】 本発明に係るキズの検出処理における回転プロジェクション法の説明に供する略線図である。
【図８】 本発明に係るキズの再現方法の説明に供する略線図である。
【図９】 本発明に係るキズの再現方法の説明に供する略線図である。
【図１０】 本発明に係るヨゴレの再現方法の説明に供する略線図である。
【図１１】 本発明に係るヨゴレの再現方法の説明に供する略線図である。
【図１２】 本発明に係るヨゴレの再現方法の説明に供する略線図である。
【図１３】 本発明に係るヨゴレの再現方法の説明に供する略線図である。
【図１４】 本発明に係るキズの種類の説明に供する略線図である。
【図１５】 本発明に係るキズの検出及び不良の判定の基準を示す略線図である。
【図１６】 本発明に係る表面情報処理装置を用いたシステムの全体構成を示すブロック図である。
【図１７】 本発明に係る不良発生工程特定処理手順を示すフローチャートである。
【図１８】 本発明に係る表面情報の重畳結果を示す略線図である。
【図１９】 本発明に係る表面情報及び工程情報を示す略線図である。
【図２０】 本発明に係る不良発生工程特定処理手順の他の実施形態を示すフローチャートである。
【図２１】 本発明に係るＬＰＤマップの重畳結果を示す略線図である。
【符号の説明】
１０ 表面検査システム
１１ レーザ散乱検出機
１２ 制御用コンピュータ
１５ ＬＡＮ
２１ 判定用コンピュータ
２５ プリンタ
３０ ＬＰＤマップ
３１ 散乱点
３２ 線状集合領域
４１ バス
４２ ＣＰＵ
４３、４９、５０ インターフェイス
４４ ＲＯＭ
４５ ＲＡＭ
４６ 表示処理部
４７ モニタ
４８ ハードディスク装置
１２１ 支援装置[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a wafer surface information processing apparatus for accumulating wafer surface information such as flaw information and scratch information on a wafer surface detected by a wafer surface inspection apparatus and using it in a wafer processing process and other processes.
[0002]
[Prior art]
In the silicon wafer manufacturing process, a lapping process using a polishing liquid containing a fine particle size abrasive is applied to a silicon wafer cut out from a silicon ingot pulled up by a CZ (Czochralski) method or the like. As a result, the surface is mirror-finished.
[0003]
The silicon wafer thus surface-finished is subjected to a surface inspection by an operator after the cleaning process, and only those that are judged to be free from scratches and dirt are shipped as non-defective products.
[0004]
[Problems to be solved by the invention]
By the way, human surface inspection has not been standardized in terms of methods and criteria, and is insufficient in terms of improving customer satisfaction.
[0005]
The present invention has been made in view of the above problems, and an object of the present invention is to provide a wafer surface information processing apparatus that accumulates wafer surface information and uses it in wafer processing steps and other steps. .
[0006]
[Means for Solving the Problems]
In order to achieve the above object, in the wafer surface information processing apparatus according to the present invention, the wafer surface information such as scratch information and scratch information on the wafer surface detected by the wafer surface inspection apparatus is used as image information and numerical values. It is characterized in that it is possible to easily detect the tendency of occurrence of scratches and scratches in a predetermined process by accumulating the information and superimposing the accumulated information.
[0007]
More specifically, the following wafer surface information processing apparatus is provided in the present invention.
[0008]
(1) Input means for fetching wafer surface information (scratch information and scratch information on the wafer surface) for each wafer supplied from the wafer surface inspection apparatus, and storage for storing the wafer surface information for each wafer for a plurality of wafers. Means, superimposing means for superimposing arbitrary wafer surface information stored in the storage means to form superimposed surface information, display means for displaying the superimposed surface information formed by the superimposing means, and various information A wafer surface information processing apparatus comprising: information processing means for processing.
[0009]
The “wafer surface inspection apparatus” detects scattered light obtained from the wafer surface when the wafer surface is irradiated with laser light, and a commercially available one can be used. The main function of this inspection apparatus is to detect each scattering point (LPD: Light Point Defect) as an individual defect.
[0010]
The “LPD aggregate (scratches) on the wafer surface” may be continuous or discontinuous. Further, the arrangement may be a straight line or a curved line. “Scratches” means various forms such as an assembly of defects on the wafer surface or rubbing scratches on the wafer surface.
[0011]
(2) The wafer surface information processing apparatus according to (1), wherein the wafer surface information and the superimposed surface information for each wafer are displayed as images on the wafer.
[0012]
(3) In the wafer surface information processing apparatus described in (1) or (2), the wafer surface information is stored in correspondence with the wafer history information in which the process passed through each wafer is recorded, and the wafer history information is stored in the wafer history information. A wafer surface information processing apparatus capable of assisting in determining in which process scratches or smudges are attached by outputting corresponding wafer surface information.
[0013]
(4) In the wafer surface information processing apparatus described in (1) or (2), the tendency of the contents of defect occurrence for each process through which each wafer passes is recorded in advance, and the wafer surface that matches the tendency of the contents of defect occurrence A wafer surface information processing apparatus characterized in that, by outputting information, it is possible to assist in determining in which process scratches and dirt are attached.
[0014]
(5) A support device for assisting in the determination of the cause of the occurrence of scratches and scratches in the wafer manufacturing process, comprising an input means, a storage means, a processing means, and an output means, and claims 3. The wafer surface information processing apparatus according to claim 3, wherein the “wafer surface information corresponding to the wafer history information” is extracted via the input means, and the extracted “wafer surface information corresponding to the wafer history information” is stored in the storage means. , A process that frequently causes scratches and scratches on the basis of the accumulated “wafer surface information corresponding to the wafer history information”, and outputs the result from the output means .
[0015]
“Scratches and scratches occur frequently” means absolute evaluation such as more than a predetermined number of times, relative evaluation that there are more than other places, the number of occurrences is the same as others, but dark or severe Includes all of the substantive evaluations such as what happens.
[0016]
(6) A support device for supporting the determination of the amount of insurance at the time of carrying the wafer, comprising an input means, a storage means, a processing means, and an output means, and the description in (3) The “wafer surface information corresponding to the wafer history information” of the wafer surface information processing apparatus is extracted via the input means, and the extracted “wafer surface information corresponding to the wafer history information” is stored in the storage means. And a support device that calculates the amount of the insurance amount based on the accumulated “wafer surface information corresponding to the wafer history information” and outputs the result from the output unit.
[0017]
(7) A computer-readable program storing a program including a superimposition process for superimposing wafer surface information (scratch information and scratch information on the wafer surface) for each wafer supplied from the wafer surface inspection apparatus to form superimposed surface information. Storage medium.
[0018]
(8) A computer-readable storage medium storing superposed surface information formed by superimposing wafer surface information (scratch information and scratch information on the wafer surface) for each wafer supplied from the wafer surface inspection apparatus.
[0019]
(9) A computer storing wafer surface information (scratch information and scratch information on the wafer surface) for each wafer supplied from the wafer surface inspection apparatus and superimposed surface information formed by superimposing the wafer surface information A readable storage medium.
[0020]
(10) The storage medium according to (8) or (9), which is a database shared among a plurality of installed apparatuses according to any one of (1) to (6).
[0021]
(11) The storage medium according to any one of (8) to (10), wherein a requirement level for a wafer of a specific wafer consumer is further stored.
[0022]
As the “storage medium”, FD, MD, HD, and the like can be cited as typical ones, but any device that can store data can be used regardless of whether it is portable or stationary. In addition, static storage is not limited to dynamic storage, and any form can be adopted. The “specific wafer consumer” means a specific customer.
[0023]
DETAILED DESCRIPTION OF THE INVENTION
Hereinafter, a surface inspection apparatus and an inspection method according to the present invention will be described with reference to the drawings.
[0024]
[Device configuration]
FIG. 1 is a schematic diagram showing the overall configuration of a silicon wafer surface inspection system 10 according to the present invention. As shown in FIG. 1, a surface inspection system 10 includes a laser scattering detector 11 for extracting minute defects from the surface of a silicon wafer, and a control computer 12 for controlling the laser scattering detector 11. And a determination computer 21 that collects extraction results (LPD maps) from these extraction units 13 via the network 15 and determines the state of scratches or scratches.
[0025]
Each laser scattering detector 11 of the extraction unit 13 irradiates the surface of the silicon wafer with laser light and detects scattered light generated when there is a defect. Then, the laser scattering detector 11 sets a map (LPD map: Light Point Defect map) in which each detected scattering point is an individual defect in a coordinate value in a predetermined coordinate system on the wafer and the intensity of the scattered light. By generating as a set, fine defects on the surface of the silicon wafer are extracted.
[0026]
Information (LPD map) of the individual scattering points (defects) on the surface of the silicon wafer extracted in this way is sent to a hard disk (not shown) provided in each laser scattering detector 11 or via the network 15. Then, it is stored in the hard disk of the direct determination computer 21 in association with the wafer ID, slot number, and the like.
[0027]
The determination computer 21 is a LAN in which the wafer ID and slot number of the silicon wafer to be determined from each control computer 12 and the scattering point information (LPD map) associated with the wafer ID and slot number are constructed by Ethernet. Obtained from the corresponding laser scattering detector 11 via (Local Area Network) (network 15).
[0028]
Incidentally, even when the plurality of control computers 12 and the laser scattering detector 11 manage LPD maps in different data formats, the determination computer 21 that has acquired these information shares these information in common. It is designed to be converted to a data format.
[0029]
Further, a computer-aided support device 121 for detecting a defect that is constantly generated by superimposing surface information (described later) of a wafer representing scratches and scratches is connected to the determination computer 21.
[0030]
Here, FIG. 2 is a block diagram showing a configuration of the determination computer 21, and a CPU (Central Processing Unit) 42, a ROM (Read Only Memory) 44, a RAM (Random Access Memory) 45, and a hard disk drive device via a bus 41. 48, a display processing unit 46, and interfaces 43, 49 and 50 are connected.
[0031]
The CPU 42 operates in accordance with a program stored in the ROM 44 or a program read from another storage medium, and the scattering point information (LPD map) of each silicon wafer supplied from the laser scattering detector 11 and the control computer 12. This is received via the interface 43 and stored in the hard disk of the hard disk drive 48.
[0032]
The CPU 42 writes a predetermined portion of the LPD map stored in the hard disk into the RAM 45 as necessary, and performs a scratch and scratch extraction process and a pass / fail determination process, which will be described later. The processing result is visually displayed on a monitor 47 such as a CRT (Cathode Ray Tube) after being subjected to graphics processing in the display processing unit 46 and, if necessary, a printer via an interface 49 such as a USB terminal. Supplied and printed.
[0033]
Here, FIG. 3 is a display example in which the LPD map 30 supplied from the laser scattering detector 11 to the determination computer 21 is converted into screen information and displayed. The determination computer 21 is generated on the surface of the silicon wafer. Of the plurality of scattering points 31, scratches or scratches that can be defective are detected in accordance with the aggregated state (feature amount). For example, among a plurality of scattering points (LPD) 31 shown in FIG. 3, a set region 32 of a series of scattering points approaching a line is detected as a flaw and an irregular set of scattering points having a high density. The region 33 is detected as a stain.
[0034]
That is, the determination computer 21 detects linear flaws and irregular irregularities from the uneven distribution of the scattering points 31 on the surface of the silicon wafer and the generation state of the aggregate according to the inspection processing procedure shown in FIG. Based on the uneven distribution of the scattering points 31 and the state of occurrence (features) of the aggregate, it is further determined whether or not the scratches or scratches should be defective. When the LPD map data is acquired from the scatter detector 11, the inspection processing procedure is entered from step A11. In step A12, the linear collection region 32 of the scattering points 31 on the silicon wafer surface is obtained from the LPD map by the section analysis method and the pyramid process. This is detected and recognized as a scratch.
[0035]
When the scratch detection process in step A12 is completed, the determination computer 21 proceeds to the subsequent step A13, and detects uneven distribution that is an irregular set of scattering points from the LPD map, that is, a stain on the surface of the silicon wafer.
[0036]
If scratches and scratches are detected, the determination computer 21 proceeds to step A14, classifies the detected scratches and scratches based on their feature values, and sets a determination criterion based on the classified results. It is used to determine whether or not the silicon wafer to be inspected at this time should be defective. The determination result is output to a printer or the like in step A15, and then the inspection processing procedure is completed in step A16.
[0037]
Here, the details of the scratch detection process in step A12 will be described. FIG. 5 is a schematic diagram for explaining the pyramid process when detecting the uneven distribution of the scattering points 31 that may be defective from the LPD map 30 and the gathering region. The determination computer 21 stores the LPD map 30 written in the RAM 45. On the other hand, each resolution is used while using three levels of resolution (a: 200 × 200 [Dots], b: 400 × 400 [Dots], c: 1000 × 1000 [Dots]) as shown in FIG. In Fig. 5, a linear collection region of scattering points is extracted by section analysis processing using a rotation projection method described later.
[0038]
That is, the determination computer 21 first analyzes the image of the LPD map 30 with a low resolution (a: 200 × 200 [Dots]) as shown in FIG. 5B, and the uneven distribution 33 of the scattering points 31 or the linear collection region. 34 is extracted. In this case, the linear collection region 34 of scattering points that can be extracted with low resolution is detected by the rotation projection method described later with this resolution.
[0039]
When the detection process at the low resolution is completed, the computer for determination 21 performs image analysis at a slightly high medium resolution (b: 400 × 400 [Dots]) as shown in FIG. Extract a linear collection area. In this case, a linear collection region of scattering points that can be extracted with this resolution is detected as a scratch by the rotation projection method described later with this resolution. Incidentally, in the case of the LPD map 30 shown in FIG. 5C, the uneven distribution 33 of the scattering points is recognized, but the linear collection region is not recognized. In such a case, there is a possibility that a linear collection region of scattering points that can be extracted with higher resolution exists in the uneven distribution 33, and the determination computer 21 further includes a determination point as shown in FIG. The LPD map 30 is image-analyzed with high resolution (c: 1000 × 1000 [Dots]), and the uneven distribution 33 and the linear gathering region are extracted.
[0040]
In the LPD map 30 with high resolution shown in FIG. 5D, a linear aggregated region 33A hidden in the uneven distribution 33, which is a high-density region of the scattering points 31, is recognized at low resolution. The computer 21 can detect the collective region 33A for the first time by this rotational projection method at this resolution.
[0041]
Next, a method for detecting a linear collection region of scattering points by the interval analysis method using the rotation projection method executed by the determination computer 21 at each resolution will be described. FIG. 6 is a schematic diagram showing the principle of a method of detecting a linear gathering region of scattering points that can be a linear defective scratch, and is a predetermined region AR10 of the LPD map 30 created based on the scattering points 31 on the surface of the silicon wafer. The segment SEG10 is cut out from this, and this segment SEG10 is rotated. Incidentally, as a method of rotating the segment SEG10, for the image data of the area AR10 extracted in the RAM 45 (FIG. 2), a method of changing the read address of the image data in accordance with the rotation angle, and the like is used.
[0042]
The projection result of each scattering point 31 on the vertical axis (Y axis) of the segment SEG10 at that time is the Y axis projection curve YP, and each scattering point 31 on the horizontal axis (X axis) of the segment SEG10 at that time Let the projection result be an X-axis projection curve XP.
[0043]
These X-axis projection curve XP and Y-axis projection curve YP have larger values as the projection amount (the number of scattering points) of the scattering points onto the respective axes (X-axis or Y-axis) increases. 6, in the segment SEG10 at a rotation angle of 50 °, the X direction of the X-axis of the segment SEG10 and the longitudinal direction of the continuous gathering region 32 among the scattering points existing inside the segment SEG10 The value of the axial projection curve XP will be partially increased.
[0044]
Therefore, it can be seen that when the X-axis projection curve XP and the Y-axis projection curve YP show a partially steep rising state, a continuous scattering point collection region 32 exists. This means that the scattering point collection region 32 is detected even if it is not continuous, and it is possible to detect a state in which scattering points are arranged with regularity in at least a certain direction in the segment SEG10. it can. Incidentally, FIG. 6 shows the states where the rotation angles are 0 °, 10 °, and 50 °. However, in the detection method of the scattering point aggregation region 32 using the rotation projection method in the present embodiment, the rotation projection is performed. The direction of the direction is not determined, but the gathering region 32 of the scattering points is detected regardless of the direction by monitoring the rising state of the X-axis projection curve XP and the Y-axis projection curve YP when continuously rotated. can do.
[0045]
Here, the detection method of the continuous linear gathering region 32 of the scattering points by the projection curve described above with reference to FIG. 6 shows the detection principle, but actually, as shown in FIG. With respect to the projection curve XP and the Y-axis projection curve YP, histograms are arranged in correspondence with the rotation angles, and the linear collection region 32 of scattering points can be determined by searching for the peak PEAK. This determination method uses a so-called two-dimensional Hough transform straight line detection method.
[0046]
Thus, the determination computer 21 uses the rotational projection method described above with reference to FIGS. 6 and 7 for each resolution described above with reference to FIG. Detect in region. The determination computer 21 recognizes a linear collection region of scattering points whose positions (segments SEG10) are specified by the rotation projection method as linear flaws by image processing based on the luminance and color, and the image information. Numerical information representing the position, shape, etc. of the object recognized as a scratch is held as wafer surface information.
[0047]
Here, when the three blocks 32a, 32b, and 32c are recognized by the CPU 42 as the linear scratch recognized in this way, for example, as shown in FIG. 6, the three blocks 32a, 32b, and 32c are set to 1 It is necessary to determine whether the book should be scratched or multiple (two or three scratches). Therefore, the determination computer 21 executes a process of connecting the linearly recognized flaws under a certain condition in the above-described step A12 (FIG. 4).
[0048]
That is, as shown in FIG. 8, as a method for determining whether or not the first block 32 a and the second block 32 b recognized as linear scratches are viewed as one scratch, the determination computer 21 includes: Using the angles θ1 and θ2 of the two scratched blocks 32a and 32b with respect to the reference direction and the angle θ3 of the straight line L1 connecting the midpoints M1 and M2 of the blocks 32a and 32b with respect to the reference direction, the angle similarity Z is obtained by the following equation.
[0049]
[Expression 1]

[0050]
This equation is to obtain the inner product of the differences between the angles. The closer the angle similarity Z is to 1, the higher the degree of connection between the two blocks 32a and 32b, that is, the degree of connection.
[0051]
This is because the angles θ1 and θ2 of the two blocks 32a and 32b are approximate, and the two blocks 32a and 32b are not separated in the perpendicular direction (the angle θ3 is small). This means that it is determined that the blocks 32a and 32b should be connected. However, even if it is determined that the degree of connection is high in this expression, if the interval between the two blocks 32a and 32b is large, these should not be connected, and the determination computer 21 is shown in FIG. It is determined whether or not the two blocks 32a and 32b should be connected according to the method.
[0052]
That is, in FIG. 9, the determination computer 21 draws a common approximate line L2 between the two blocks 32a and 32b, and a predetermined value in which the interval L3 in the direction of the approximate line L2 between the two blocks 32a and 32b is set in advance. If smaller, it is determined that the two blocks 32a and 32b should be connected.
[0053]
Thus, by determining the necessity of connection based on the distance L3 in the direction of the approximate straight line L2 between the two blocks 32a and 32b, the determination is made based on the distance L4 between the two closest points of the blocks 32a and 32b. Compared to the case, the two adjacent blocks 32a and 32b to be connected can be reliably connected.
[0054]
Next, a detailed processing procedure in the detection processing step A13 for the uneven distribution (dirt) of the scattering point 31 on the surface of the silicon wafer shown in FIG. 4 will be described. FIG. 10 shows an LPD map 30 showing a detection state of the scattering point 31 on the surface of the silicon wafer, and shows a state where unevenly distributed regions 35a, 36a, and 37a of the scattering point 31 exist. In this state, the scattering points 31 of the LPD map 30 are represented as dots as shown in FIG. For each dot of the LPD map 30, the determination computer 21 performs conversion processing into a bitmap of 256 gradations, for example, to obtain bitmap data BM as shown in FIG.
[0055]
The determination computer 21 performs a smoothing process using a spatial filter on the bitmap data BM, thereby obtaining a smoothing curve S1 obtained by smoothing the bitmap data BM as shown in FIG. When the smoothing curve S1 is represented as an image, only the unevenly distributed areas 35a, 36a, and 37a of the scattering point 31 are represented in a blurred state as shown in FIG. In this display state, as a result of the smoothing process, the dots that are farther from the other dots are displayed lighter.
[0056]
Then, as shown in FIG. 11C, the determination computer 21 binarizes the smoothing curve S1 with a preset threshold SH to obtain binarized area data D35C. By the binarized area data D35C, unevenly-distributed areas 35c, 36c, and 37c that are clearly different from the surroundings as shown in FIG. 10C are obtained. In this way, by extracting the threshold SH for the unevenly distributed region of the scattering points 31, even if the density of the scattered points is slightly higher than that of the surroundings, the unevenly distributed region is reliably determined by selecting the threshold SH. Can be extracted.
[0057]
In addition, by appropriately changing the threshold value SH when obtaining the binarized data D35C as described above, the unevenly distributed region can be detected according to the detection state of the scattering points 31 on the surface of the silicon wafer (the overall density of the scattering points 31). It can be carried out. For example, when a large number of scattered points 31 are detected as a whole, by increasing the setting level of the threshold SH, a region with a particularly high detection density is defined as an unevenly distributed region, and other regions where the scattered points 31 are present on average. It becomes possible to detect them in distinction. The determination computer 21 recognizes the irregularly-distributed uneven area detected in this way as a stain, and holds this as numerical information (wafer surface information) representing image information, position, shape, and the like.
[0058]
When scratches and scratches on the surface of the silicon wafer are detected as described above, the determination computer 21 determines whether or not these detected scratches and scratches should be defective in processing step A14 shown in FIG. Judgment in
[0059]
That is, the determination computer 21 determines whether or not the linear flaw extracted in step A12 in FIG. 4 is defective based on the length and detected intensity. In this case, the length of the scratch means the length of the scratch recognized as one by the method described above with reference to FIGS. 8 and 9, and the detected strength of the scratch means the rotation projection described above with reference to FIGS. This means the peak value PEAK (corresponding to the density of scratches) of the projection curves XP, YP (histogram) in the method.
[0060]
Further, the determination computer 21 obtains a circumscribed rectangle with respect to the uneven distribution (dirt) extracted in step A13 in FIG. 4, and determines the area of the circumscribed rectangle, the length in the vertical direction, the length of the horizontal method, the length of the diagonal line, Whether or not it is defective is determined based on the density and area-density determination curve. That is, as shown in FIG. 12, the computer for determination 21 obtains a circumscribed square 35d of the unevenly distributed (going) area 35c extracted in the above step A13, and the vertical length H and the lateral direction of the circumscribed square 35d are obtained. The length W and the diagonal length D and the area are measured, and the density is obtained on the basis of the integrated value of each scattering point 31 in the unevenly distributed region 35c.
[0061]
Based on the area and density of the circumscribed rectangle 35d, the determination computer 21 determines that the defect is satisfied when the area-density determination formula expressed by (density−asymptotic density)> coefficient α / (area−asymptotic area) is satisfied. . This means that a defect is determined when the density and / or area values are larger than the area-density determination curve S35 as shown in FIG. That is, focusing on the fact that the area and density are in an inversely proportional relationship as a condition for determining that a person is obscured, what is expressed as a conditional expression is the above-described area-density determination expression. By using this conditional expression, it is possible to perform the determination of a defect based on the same condition, which has conventionally been determined subjectively by a person.
[0062]
In addition to the basic determination method for defects based on scratches and scratches, the determination computer 21 classifies the types of scratches and scratches according to the feature amount of the scattering point collection area and the uneven distribution area, and according to the classification result. In this way, the criterion for defect determination is set. That is, as shown in FIG. 14, the scratches formed on the surface of the silicon wafer include, for example, scratches in a regular arc-shaped arrangement (hereinafter referred to as first scratches) 39 generated in the lapping process, There are irregularly arranged flaws (hereinafter referred to as second flaws) 40 consisting of a collection of defects called pock marks accompanied by deterioration of pressure resistance (GOP), which is more second than the first flaw 39. It is necessary to judge the scratch 40 as a serious defect.
[0063]
Therefore, in the determination processing step A14 shown in FIG. 4, the determination computer 21 determines the type of scratch according to the set state (feature amount) of the scattering points, and changes the determination criterion according to the type of scratch. Has been made. For example, with respect to a second scratch having a relatively irregular arrangement of scattering points as a feature quantity, the determination criterion is compared to the case of the first scratch having a regular arrangement of scattering points as a feature quantity. By shortening the length, the second flaw is defective even if the length is not defective in the first flaw.
[0064]
In addition, in the determination processing step A14 shown in FIG. 4, the determination computer 21 determines the type of blur according to the set state (features) of the scattering points described above with reference to FIG. 12, and according to the type of blur. The determination criterion (area-density determination curve S35) described above with reference to FIG. 13 is changed.
[0065]
Incidentally, FIG. 15A is a diagram showing a method for detecting scratches and blurs by the judgment computer 21, and FIG. 15B is a diagram showing judgment criteria for defective wafers. Here, the length, darkness, area, and density are used as feature quantities for classifying scratches and scratches. However, the length, density, width, linearity, circularity, There are positions, etc., and the feature values of the stain are its area, density / density, distribution, shape, position, etc., and the determination computer 21 uses these feature values as needed.
[0066]
Thus, the determination computer 21 determines a defect based on scratches or scratches on the silicon wafer surface, and the result is output by a printer or the like.
[0067]
The surface inspection system 10 having the determination computer 21 has a function of specifying a process in which scratches and scratches have occurred, in addition to such scratch and scratch detection (wafer surface information generation) and defective wafer determination processing. Yes.
[0068]
In other words, FIG. 16 in which the same reference numerals are assigned to the corresponding parts in FIG. 1 shows each extraction unit 13 (13A, 13B, 13C) having a laser scattering detector (wafer surface inspection apparatus) 11 (11A, 11B, 11C). 2 is a block diagram showing an example of a configuration in which each of the manufacturing steps of the wafer 100 is arranged before and after each step (for example, a first step as a surface polishing step and a second step as a cleaning step).
[0069]
In FIG. 16, the LPD maps LPD1, LPD2, and LPD3 detected by the laser scattering detectors 11A, 11B, and 11C are stored in the hard disk of the determination computer 21 in association with each process information. That is, each LPD map LPD1, LPD2, and LPD3 output for each wafer from each laser scattering detector 11A, 11B, and 11C is accompanied by the wafer ID, slot number, etc. of the LPD map. 21 designates information for identifying these wafers and, in addition to this, information for identifying the laser scattering detectors 11A, 11B or 11C which are the output sources of the LPD map (IDs of the laser scattering detectors 11A, 11B and 11C, etc.) I remember it with the map.
[0070]
Information for specifying the laser scattering detector that is the output source of the LPD map stored in the determination computer 21 is associated with the wafer ID, slot number, etc. of the wafer corresponding to the information in time series. Going to accumulate. Therefore, the determination computer 21 confirms the process history of the wafer and the final process at the current time point based on the accumulated ID of the laser scattering detector for the wafer (the accumulated information is called process history information). Can do.
[0071]
Therefore, for example, the process history information corresponding to the LPD map LPD1 output from the laser scattering detector 11A provided before the first process is in a state in which no information is accumulated. Thus, it can be recognized that the LPD map LPD1 is an LPD map of a certain wafer before the first step, and corresponds to the LPD map LPD2 output from the laser scattering detector 11B provided before the second step. The process history information to be performed is a state in which only the information representing the first laser scattering detector 11A is accumulated, and the computer 21 for determination therefore indicates that the LPD map LPD2 is at the end of the first process, that is, before the second process. It can be recognized that this indicates an LPD map of a certain wafer.
[0072]
Incidentally, as the process information, instead of accumulating and using the ID of each laser scattering detector 11, as shown in FIG. 16, the wafer ID and the slot number of the wafer processed in each process are shown together with information indicating the process. The process information D (PRO) may be supplied to the determination computer 21 via the network 15. In this case, the determination computer 21 can recognize which process has been completed for the wafer specified by the wafer ID and the process history of the wafer based on the wafer ID supplied from each process.
[0073]
Thus, each LPD map supplied from each laser scattering detector 11 is stored in the determination computer 21 in association with the process history information of each wafer. The accumulated LPD map is used for detection of scratches and scratches by the processing steps described above with reference to FIG.
[0074]
Scratches and scratches information (wafer surface information) detected in the processing procedure of FIG. 4 are determined as image information shown in FIGS. 14 and 10C and numerical information such as the position, size, and darkness thereof. The determination computer 21 superimposes the surface information on wafers having the same process history. The superimposed result is supplied to the support device 121, and is used when specifying a process in which scratches or scratches have occurred.
[0075]
That is, FIG. 17 shows a specific process procedure of a defective process by the determination computer 21 and the support device 121. When the determination computer 21 enters the process procedure from step A21, in step A22, the scratches and scratches described above with reference to FIG. The wafer surface information already generated in the detection process is superimposed on the wafer information of the same process information.
[0076]
As a result, as shown in FIG. 18, scratches occurring at the same position are overlapped at the same position, so that scratch information and dirt information in a certain range (areas 140, 141, 130, 131 in FIG. 18) are obtained. When the occurrence frequency of the wafer becomes high, this means that the wafer in any one of the process histories of the wafers on which the surface information is superimposed (the process histories of the respective wafers on which the surface information is superimposed is the same). This means that scratches and scratches that cause defects are constantly occurring.
[0077]
Therefore, the computer for determination 21 displays the superimposition result as an image on the monitor in step A23, and then supplies the superimposition data to the support apparatus 121, and the support apparatus 121 executes the defective process specifying process in step A24.
[0078]
In this specifying process, the support apparatus 121 uses superimposition data (for example, superimposition data Dn and Dm shown in FIG. 19) in which the final process differs by one process among the superimposition data in which the surface information corresponding to each process history information is superimposed. Compare. For example, as shown in FIG. 19, superimposition data Dn (that is, the final process is that of the first process) obtained by superimposing the surface information D1, D2,. There is a difference in whether or not the second process is included in the superimposition data Dm obtained by superimposing the surface information D11, D12,. Therefore, the support device 121 compares these superposition data, and when the frequently occurring scratch or blur is not in one of the superimposition data but in the other superimposition data, the cause of the occurrence of the scratch or blur is in the second step. It can be determined that there is.
[0079]
The support apparatus 121 stores the determination result and the superimposition result in a storage medium such as a hard disk and outputs them to a monitor or printer, and then ends the processing procedure in step A25.
[0080]
By the way, instead of using superimposition data used for the determination as described above for different wafers, the same wafer superimposition data is stored every time the process history is added, and the stored same wafer is stored. You may make it compare the superimposition data about.
[0081]
[Operation]
In the silicon wafer surface information processing apparatus (determination computer 21 and support apparatus 121) according to the present invention having the functions and configurations described above, scratches and scratches are generated based on the LPD map supplied from the laser scattering detector 11. Is generated as information consisting of image information and various numerical values, and the generated surface information is superimposed on a plurality of wafers. In this case, if there is a tendency for many scratches and scratches to occur at a certain position on the wafer surface, the overlapping amount of similar scratches and scratches at that position increases as a result of the superposition.
[0082]
Therefore, it is possible to determine whether the scratches or scratches detected for each wafer are accidentally generated only on the wafer or are generated constantly by looking at the superimposed result. It becomes possible.
[0083]
The process in which scratches and scratches occur can be identified by the process history information associated with the surface information of each wafer, and the support device 121 identifies the process in which scratches and scratches have occurred based on the process history information. To do. Therefore, the operator can easily find the defect generation process only by looking at the specific result. Then, the operator can improve the work process by quickly removing the continuous generation of defective products by improving the inspection of the found defect generation process.
[0084]
[Other embodiments]
In the above-described embodiment, as described above with reference to FIG. 17, scratches and scratches detected based on the LPD map in the determination computer 21 are generated as image information and various numerical information, and the generated information ( However, the present invention is not limited to this, and the LPD map supplied from the laser scattering detector 11 is used as the wafer surface information to determine the computer 21 and the support device 121. May be superimposed. The processing procedure in this case is shown in FIG.
[0085]
That is, FIG. 20 shows another embodiment of a process procedure for identifying a defective process by the determination computer 21 and the support device 121. When the determination computer 21 enters the process procedure from step A31, in FIG. The LPD map supplied from the laser scattering detector 11 as described above is superimposed on the wafers having the same process information.
[0086]
As a result, as shown in FIG. 21, the LPD occurring at the same position is superimposed at the same position, so that the frequency of occurrence of LPD in a certain range (areas 240, 241, 230, and 231 in FIG. 21) is increased. If it becomes higher, this means that scratches or scratches that make the wafer defective are constantly occurring in any of the process histories of the respective wafers on which the LPD map is superimposed. . Incidentally, the method described above with reference to FIGS. 1 to 15 is used as a method for detecting scratches and scratches based on the superimposed LPD in this case.
[0087]
The computer for determination 21 displays the superimposition result and the detection result as an image on the monitor in step A33, and then supplies the superimposition data to the support apparatus 121. The support apparatus 121 executes the defect process specifying process in step A34. This defective process specifying process is the same as the process step A24 described above with reference to FIG.
[0088]
Thus, the computer for determination 21 and the support apparatus 121 can specify the process that is the cause of the occurrence of scratches and scratches by directly superimposing the LPD map. In this way, when the LPD map is directly superimposed, in addition to defects due to external factors such as scratches and scratches, small LPDs that are not considered scratches or scratches are constantly generated at the same position on each wafer. Such a defect (region 242 shown in FIG. 21) can also be detected.
[0089]
Further, in the above-described embodiment, the case where the process performed by the wafer is recorded as the process history information has been described. However, the present invention is not limited to this, and various methods such as recording only the final process at the present time are described. Can be applied.
[0090]
In the above-described embodiment, the case where the determination computer 21 superimposes the wafer surface information and supplies the superimposed data to the support device 121 to identify the defect generation process has been described. However, the present invention is not limited to this. Scratch information and blur information (wafer surface information) detected and generated by the determination computer 21 may be supplied to the support device 121, and the support device 121 may superimpose the surface information.
[0091]
In the above-described embodiment, the surface polishing process and the cleaning process are described as examples of the wafer manufacturing process. However, the present invention is not limited to this, and the present invention may be applied to, for example, a wafer transfer process. Can do. In this way, it is possible to determine the occurrence of scratches or scratches during conveyance. In this case, the support device 121 obtains process history information and wafer surface information corresponding to the process history information from the computer for determination, accumulates them in a storage medium such as an internal hard disk, and insures during wafer transfer based on the accumulated information. It becomes possible to judge the amount of money.
[0092]
In the above-described embodiment, the system having one set of the determination computer 21 and the support device 121 has been described. However, the present invention is not limited to this, and each surface information is determined by the plurality of determination computers 21 and the support devices 121. Alternatively, the superimposed data may be shared.
[0093]
In the above-described embodiment, a system for detecting scratches and scratches and identifying a process in which these have occurred has been described. However, the present invention is not limited to this. For example, in the determination computer 21 or the support device 121, a consumer The required quality for the wafer is held, the wafer whose quality is judged to be deteriorated from the required quality is determined, and the process in which the defect occurring in the wafer is identified is specified. Good. In this case, the quality is determined on the basis of the shape and density of scratches and scratches in the determination process described above with reference to FIG.
[0094]
In the above-described embodiment, the laser scattering detectors 11A, 11B, and 11C are provided between the plurality of steps, respectively, and the LPD maps obtained from the plurality of laser scattering detectors 11A, 11B, and 11C (that is, between the steps). Although the case where a process in which a defect occurs is specified based on a plurality of detected LPD maps), the present invention is not limited to this. For example, a laser provided after the final process of all processes or a predetermined process unit You may make it judge a defect generation process only from the LPD map obtained from the scattering detector 11C.
[0095]
In this case, in the determination computer 21 or the support device 121, the tendency of the defect contents (scratches and scratches, radius of curved scratches, etc.) generated in each process is stored in advance, and the LPD detected after the final process is stored. The map is superimposed in the same manner as described above with reference to FIG. 17 or FIG. 19, and information on scratches and scratches (wafer surface information) is detected based on the result of the superimposition. And the process which has such a defect generation tendency is specified based on what matches the tendency of the defect content memorize | stored beforehand among the surface information of the wafer detected in this way. In this way, it is possible to specify a defect occurrence process without providing a laser scattering detector between each process. Incidentally, the information indicating the tendency of occurrence of defects for each process includes the items described above as feature values of scratches and scratches (the feature values of scratches include length, density, width, linearity, circularity, position, etc.). In addition, as the feature amount of the stain, its area, density / density, distribution, shape, position, etc.) can be used.
[0096]
【The invention's effect】
As described above, the silicon wafer surface information processing apparatus according to the present invention can detect flaws and scratches that occur constantly by superimposing the wafer surface information, and the occurrence thereof. It is also possible to detect the process. Accordingly, it is possible to improve the inspection efficiency.
[Brief description of the drawings]
FIG. 1 is a schematic diagram showing the overall configuration of a silicon wafer inspection system according to the present invention.
FIG. 2 is a block diagram showing a configuration of a defect determination computer according to the present invention.
FIG. 3 is a schematic diagram illustrating an example of detection of an LPD map.
FIG. 4 is a flowchart showing a silicon wafer surface inspection processing procedure according to the present invention.
FIG. 5 is a schematic diagram for explaining pyramid processing in scratch detection processing according to the present invention.
FIG. 6 is a schematic diagram for explaining a rotational projection method in a flaw detection process according to the present invention.
FIG. 7 is a schematic diagram for explaining a rotation projection method in a flaw detection process according to the present invention.
FIG. 8 is a schematic diagram for explaining a scratch reproduction method according to the present invention.
FIG. 9 is a schematic diagram for explaining a scratch reproduction method according to the present invention.
FIG. 10 is a schematic diagram for explaining a blur reproduction method according to the present invention.
FIG. 11 is a schematic diagram for explaining a blur reproduction method according to the present invention.
FIG. 12 is a schematic diagram for explaining a blur reproduction method according to the present invention.
FIG. 13 is a schematic diagram for explaining a blur reproduction method according to the present invention.
FIG. 14 is a schematic diagram for explaining types of scratches according to the present invention.
FIG. 15 is a schematic diagram illustrating a criterion for detecting a defect and determining a defect according to the present invention.
FIG. 16 is a block diagram showing an overall configuration of a system using the surface information processing apparatus according to the present invention.
FIG. 17 is a flowchart showing a defect occurrence process specifying process procedure according to the present invention.
FIG. 18 is a schematic diagram illustrating a superimposition result of surface information according to the present invention.
FIG. 19 is a schematic diagram showing surface information and process information according to the present invention.
FIG. 20 is a flowchart showing another embodiment of a defect occurrence process specifying process procedure according to the present invention.
FIG. 21 is a schematic diagram showing a superposition result of an LPD map according to the present invention.
[Explanation of symbols]
10 Surface inspection system
11 Laser scattering detector
12 Control computer
15 LAN
21 Computer for judgment
25 Printer
30 LPD map
31 Scattering points
32 linear assembly area
41 bus
42 CPU
43, 49, 50 interface
44 ROM
45 RAM
46 Display processing section
47 Monitor
48 Hard disk device
121 Support device

Claims (8)

Input means for fetching wafer surface information of a plurality of wafers supplied from the wafer surface inspection apparatus for each wafer each time a process is performed on the wafer;
Storage means for storing wafer surface information for each wafer by associating wafer surface information for each wafer with process history information representing a history of a plurality of processes for the wafer;
Superimposing means for superimposing image data based on a plurality of wafer surface information corresponding to the same process history in different wafers accumulated in the storage means to form superimposed surface information corresponding to each process history;
Display means for displaying superimposed surface information formed by the superimposing means;
Detecting means for detecting defects based on the superimposed surface information formed by the superimposing means;
Based on the occurrence state of the defect detected by the detection means, it is determined whether or not the defect should be defective, and it corresponds to the superimposed surface information based on the superimposed surface information corresponding to the same process history. Superimposition surface information that is different in the last process of the attached process history is compared, and there is a defect that is not included in the superimposition surface information of the front side process, but should be a defect included in the superposition surface information of the rear side process Determining means for performing a process for identifying a defective process by identifying a process to be performed;
A wafer surface inspection apparatus comprising: The judging means
Based on the defect occurrence state detected by the detection means, defects in the linear collection area are classified as flaws, and defects that are not determined as flaws are classified as scratches. In the case of (1), whether or not the wafer to be inspected should be defective is judged based on the criteria based on the length of the collecting region and the detection intensity of the defect, and in the case of blurring, on the basis of the area of the collecting region and the density of the defect. ,
When the determination means classifies the defect, the segment including the area where the defect is detected is cut out from the image data based on the superimposed surface information, the number of defects on the vertical and horizontal axes of the segment is measured, and the segment is rotated. Measure the number of defects with respect to the vertical and horizontal axes of the segment when a predetermined rotation angle is reached, and if it is detected that the amount of defects has increased, the defects are regulated in a certain direction. The linear gathering area is detected in the entire area of the wafer to identify the position of the linear gathering area, and further, the linear gathering area of the identified defect is determined. The wafer surface inspection apparatus according to claim 1, wherein the wafer surface inspection apparatus is classified as a linear flaw by performing image processing based on the luminance and the color. When the scratches classified by the judging means are recognized in a state of being divided into a plurality of blocks,
The judging means
For two flawed blocks adjacent to each other, the angles with respect to the reference direction of each block are θ1 and θ2, and the angle with respect to the reference direction of the straight line L1 connecting the midpoints M1 and M2 of each block is θ3. The obtained angle similarity Z is determined, and whether or not two adjacent blocks should be connected as one block is determined based on the angle similarity Z and the interval between two scratch blocks. The wafer surface inspection apparatus according to claim 2, wherein:

When the judgment means classifies defects,
Image analysis is performed with a predetermined resolution to detect the uneven distribution of defects and linear gathering areas. When the uneven distribution of defects is detected, the uneven distribution area of the defects is analyzed with a resolution higher than the predetermined resolution. 4. The wafer surface inspection apparatus according to claim 1, wherein uneven distribution and a linear gathering area are detected, and a linear gathering area included in the defect uneven distribution area is detected. 5.   The wafer surface inspection apparatus according to claim 1, wherein the display unit displays the wafer surface information and the superimposed surface information as an image on the wafer. The process stored in the storage means includes a wafer manufacturing process and a wafer transfer process,
6. The support unit according to claim 1, further comprising: support means for determining an insurance amount to be applied at the time of wafer transfer based on information on occurrence of defects generated in the transfer process based on wafer surface information associated with the transfer process. Wafer surface inspection equipment. The judging means
Defects are discriminated as first flaws in an arc-shaped regular arrangement, second flaws in an irregular arrangement, and scratches, and the second flaw is judged to be more severe than the first flaw. The wafer surface inspection apparatus according to claim 1, wherein it is determined whether or not the defect should be defective based on different determination criteria depending on the type of defect. A wafer surface inspection method for inspecting defects formed on the surface of a wafer,
Capturing wafer surface information of a plurality of wafers for each wafer each time a process is performed on the wafers;
Correlating wafer surface information for each wafer with process history information representing a history of a plurality of processes for the wafer, and storing the wafer surface information for each wafer;
Superimposing the accumulated image data based on a plurality of wafer surface information corresponding to the same process history in different wafers to form superimposed surface information corresponding to each process history;
Displaying the formed superimposed surface information;
Detecting defects based on the formed superimposed surface information;
Based on the occurrence state of the defect detected by the detection means, it is determined whether or not the defect should be defective, and it corresponds to the superimposed surface information based on the superimposed surface information corresponding to the same process history. Superimposition surface information that is different in the last process of the process history to be attached is compared, and a defect that should not be included in the superimposition surface information of the rear process, but not included in the superimposition surface information of the front process, is generated Performing a process of identifying a defective process by identifying a process to be performed;
A wafer surface inspection method comprising:

Images