JP3603880B2 - Overlay exposure method - Google Patents

Description

【０００９】
波長により解像度が異なってしまうことになる。この結果波長に比例して解像度が変わり、解像度の異なる各波長での波形の合成が検出されることになり、波長合成の効果が低減されることになる。
【００１０】
さらに従来波長バンド幅の広い波長で検出しようとすると、検出光学系に含まれる色収差、特に横色収差の影響を受けて例えばパタ−ンのエッジを検出しようとすると波長ごとにエッジ位置が変わり、アライメント検出や寸法検出、位置検出の誤差となってしまう。このような誤差は検出制度が０．１μｍ程度ではあまり問題にならなかったが、０．０５μｍ以下の精度が要求される今後の計測技術では大いに問題となるところである。しかしながらこのような精度を満足する検出光学系、即ち顕微鏡対物レンズや、縮小レンズの製作は困難を極めている。
【００１１】
【課題を解決するための手段】
上記の課題を解決するために本発明は以下に示す手段を用いる。光源として波長幅の広い光源もしくは実効的に複数の単波長を含む光源を用い、この光源から出射した光を、層構造から成りその表面の最上層の少なくとも一部が光学的に透明な被検物体に照明する。そしてこの被検物体で反射した光を用いてこの被検物体表面のパタ−ンを検出する分けであるが、この際この被検物体の層構造および材質の情報に応じて上記の光源から出射する光の分光照明強度特性（波長強度分布特性）を実効的に変化させ、この変化により得られた所望の分光照明強度で照射された被検物体のパタ−ン像を一次元もしくは二次元の像として検出する。
【００１２】
上に述べた検出法の更に具体的な手段としては、例えば、上記の光源から出射する光を分光手段により波長ごとに光の進行方向が変化するようにした後、分光された光路中に各波長の光の通過する場所ごとに光の透過率を異ならしめる空間光透過率変調器を配置して、上記分光照明強度特性（波長強度分布特性）を実効的に変化させればよい。
【００１３】
また別の具体例として、照射された被検物体のパタ−ン像を一次元もしくは二次元の像として検出する際に、被検物体で反射した光の光路中にその反射光の波長ごとに透過率を変化せしめる分光透過率可変フィルタを挿入し、検出光の分光透過率を変えることにより、最適な分光強度の照明をしていることに成り、上記光源から出射する光の分光照明強度特性（波長強度分布特性）を実効的に変化させていることになる。
【００１４】
本発明は更に前述の波長ごとに解像度が異なる従来技術の課題を解決するために、上記所望の分光照明強度で照射された被検物体のパタ−ン像を一次元もしくは二次元の像として検出する際に、被検物体で反射した反射光の各波長ごとに上記パタ−ン検出の開口数（ＮＡ）を異ならしめる。
【００１５】
更にこの際、上記各波長ごとのパタ−ン検出の開口数に対し一定の比率を持つ照明のパ−シャルコヒ−レンスになるように照明光の各波長ごとに指向性を持たせている。
【００１６】
上記の色収差に対する課題を解決するため、本発明では波長幅の広い光源もしくは実効的に複数の単波長を含む光源から出射した光を、被検物体に照明し、被検物体で反射した光を用いて該被検物体表面のパタ−ン検出する。この際、光源から出射し、被検物体で反射した光を格子等の分光手段上にほぼ結像させ、この分光手段により分光されたパタ−ンの分光方向と直交するパタ−ン検出方向については分光手段上のパタ−ンを結像し、分光方向については分光した非結像の状態で分光一次元像として二次元撮像手段により検出する。しかも分光一次元像の各分光波長に対応するパタ−ン検出方向の各信号に対し、検出光学系による色収差を信号処理により補正した後、色収差補正されたこれら各信号を合成し、この合成信号によりパタ−ン検出する。
【００１７】
上記の分光一次元像を検出する方法を用いれば、各分光波長に対応するパタ−ン検出方向の各信号に対し、各波長の信号全体に各波長ごとに異なる定数を掛けた後、各波長の信号を合成し、当該合成波長によりパタ−ン検出すれば上記の照明光の分光特性を変化させ所望の分光照明強度で照射された被検物体のパタ−ン像を検出した場合と同一の効果がある。
【００１８】
また上記分光一次元像を用いる方法を用いれば、各分光波長に対応するパタ−ン検出方向の各信号に対し、各波長の信号の場所により異なる定数を各波長ごとに掛けた後、各波長の信号を合成し、当該合成波長によりパタ−ン検出れば、各場所ごとに異なる層構造および材質に応じて、各場所ごとに最適な分光照明強度を与えて検出したことになる。
【００１９】
上記分光照明強度を所望なもの、すなわち最適なものにするため、本発明は、上記被検物体の層構造および材質の情報を用いて検出光の各波長における層構造に起因する多重干渉強度を計算し、パタ−ンの構造を忠実に反映する検出信号波形が得られる様に上記光源から出射する光の分光照明強度特性（波長強度分布特性）を被検物体の層構造および材質の情報に応じて実効的に変化させる。
【００２０】
以上に説明したパタ−ン検出手段は単なるパタ−ン検出或いはパタ−ン位置検出装置にのみ有効な手段ではなく、投影露光装置のアライメントに必要なアライメントパタ−ン検出系の手段ともなる。
【００２１】
ウエハに塗布されたレジストの下に形成されているアライメントパタ−ン、或いはこのウエハを露光後のアライメントパタ−ン、或いは現像後さらにはエッチング後のアライメントパタ−ンを上記のパタ−ン検出手段を用いてそのパタ−ン位置を正確に計測し、この結果をレジスト塗布機、露光装置、現像装置、或いはエッチング装置にフィ−ドバック、或いはフィ−ドフォワ−ドする様に露光システムを構成することにより、より高い精度で重ね露光が実現する。
【００２２】
とりわけこの露光システムでは露光装置内のパタ−ン検出系、即ちアライメントパタ−ン検出系の分光照度（水銀ランプの輝線スペクトル、或いはレ−ザ光等の単色光でも、多色光でも良い）や照明の指向性或いは検出光学系の開口数等の光学条件と同一の条件にパタ−ン位置検出系を設定し、露光装置で露光する前のアライメントパタ−ンの位置を検出し、得られたこの第一のパタ−ン位置情報と、検出しているアライメントパタ−ンに対する最適な検出光学系条件で、かつ第一のパタ−ン位置情報を得たときと同一のウエハ位置で検出し、得られた第二のパタ−ン位置情報との差を求め、この差を露光装置のアライメントオフセットとして、露光装置にフィ−ドフォワ−ドすれば、露光装置のアライメント検出系が例えウエハのアライメントパタ−ンに対し最適な条件に成っていなくとも、正しいオフセット補正が成され、精度の高い重ね露光を実現することができる。
【００２３】
【作用】
前に示した図１４を例にとって本発明の作用を説明する。下地が図１４（Ａ）のように波長により反射率が異なっていれば図１５（Ａ）に示すような分光透過率を有するフィルタ−を通して照明すれば、図１５（Ｂ）のように下地の反射率が一定で完全な白色照明をしたのと同等になり、即ち、多重干渉の条件が波長依存性がなくなり、図１５（Ｃ）に示すように検出波形は対称になり、多重干渉による検出誤差が小さくなる。
【００２４】
一般には下地の反射率は上記のような単純な反射率ではなく、特に下地が多層構造の場合複素反射率となる。このような場合でも複素反射率を下地の層構造および材質の特性（分光複素屈折率）から計算によりもとめることができるし、光学的に計測することも可能であるので、このようにして得られた複素反射率と下地の上に乗る薄膜の膜厚と屈折率から照明光の最適分光強度を求め、このような分光特性の照明を上記の手段で実現して検出すれば、精度の高い検出を行うことができる。
【００２５】
また本発明では検出波長により異なる検出光学系の開口数を各波長にたいし同一になるようにすることにより、理想光学系においても発生するパタ−ンエッジ部での色ずれ（色にじみ）がまったくなくなり、完璧な検出が可能となる。
【００２６】
また理想的でない色収差のある光学系に対しても、被検出物で反射した光を一旦格子等の分光手段上に結像後に分光した光をパタ−ンの検出方向には結像し、これと直交する方向には分光した状態で二次元撮像装置に結像することにより、各波長での像を別々に検出することが可能となる。この結果、色収差のある光学系で検出され、色ずれを持ったまま結像している像を各波長ごとに色ずれを画像処理により補正してやることが可能となる。
【００２７】
さらに各波長での像が別々に検出されているため、上記の例のように照明光の分光照度特性が所望のものになるようにハ−ド的に補正する必要がなくなる。即ち、検出された各波長の画像に上記の所望の分光特性で照明したのと同等の結果が得られるように各波長の信号にそれぞれ異なる定数を掛けあわせて得られた信号を合成すれば良い。
【００２８】
各波長での像が別々に検出できる方式を用いれば、パタ−ンの部分ごとに異なる薄膜の厚さに応じて、その部分ごとに検出光に用いる波長を選ぶことが可能になるため、最も検出誤差の小さくなる検出が実現できる。
【００２９】
【実施例】
以下本発明を実施例により更に詳細に説明する。
【００３０】
図１は本発明のパタ−ン検出方法をステッパ（投影露光装置）に適用した実施例である。レチクル３に露光照明系６からｉ線、あるいはエキシマレ−ザ光を照射し、レチクルを透過した光を縮小レンズ５によりレチクル描画パタ−ンをｘｙｚ方向に移動可能なウエハステ−ジ７上のウエハ４の露光領域４０に投影露光する。ウエハステ−ジをｘｙ方向にステップ移動することにより、図２に示すようにウエハ２上には露光された回路パタ−ンの領域が配列される。隣合う各露光回路パタ−ンの領域の間には次の露光プロセスの際に既に露光形成されている回路パタ−ンとの重ね合わせを高精度に実現するためにアライメントパタ−ン４１が形成されている。このアライメントパタ−ン４１は図３に示すように凹部４１１のある下地パタ−ンの上にフォトレジスト４１２が塗布されている。このパタ−ンを縮小レンズ５を通して精度良く、位置検出する必要がある。
【００３１】
アライメントパタ−ン４１のｘ方向の位置を検出する検出系は点線１０ｘである。図には示されていないが、ｙ方向の位置用のアライメントパタ−ンを検出する検出系１０ｙは紙面と垂直な方向にある。検出系１０ｘにはパタ−ン検出用の例えばハロゲンランプやキセノンランプ等からなる照明光源１１がある。照明光源１１を出射した光は所望のビ−ム径と指向性にするレンズ１０１を透過し、ミラ−１０２で反射後、分光強度および指向性調整照明手段１２を透過する。この分光強度および指向性調整照明手段１２は照明光を最適な分光照度分布にすると共に、それぞれの波長の光を所望の指向性となるようにする。ここを透過した光は、ビ−ムスプリッタ１０３で反射し、レンズ１０４、ミラ−１０５を通過し、縮小レンズ５を透過した後、ウエハ４上のアライメントパタ−ン４１を照明する。アライメントパタ−ン４１で反射した光は再び縮小レンズ５を通過した後、ミラ−１０５で反射し、検出系１０ｘに入る。入ってきた検出光はレンズ１０４、ビ−ムスプリッタ１０３を通過後波長選択ビ−ムスプリッタ１０６で２分される。
【００３２】
このビ−ムスプリッタは波長がλ_１からλ_２までの光を反射し、λ_２からλ_３までの光を透過する。このように検出光を２つの波長帯域に分離し、検出するのは縮小レンズ５が露光光で最高の解像度が得られるように設計、製作されているため、λ_１からλ_３までの広い波長幅の光に対し、大きな色収差を持ってしまい、これを補正する色補正レンズの製作が困難となるためである。即ち図１に示すように２つの波長帯域に分離しておけばλ_１からλ_２までの光はレンズ１０４と１０８により色補正でき、λ_２からλ_３までの光は１０４と１０８’により色補正できる。波長選択ビ−ムスプリッタ１０６で分離されたλ_１からλ_２までの光は分光透過率および開口数調整検出手段１３を透過する。この分光透過率および開口数調整検出手段１３は各検出波長に対し、最適な検出開口数で検出することを可能とする。波長選択ビ−ムスプリッタ１０６で分離さえた他方の波長がλ_２からλ_３までの光はミラ−１０７で反射され分光透過率および開口数調整検出手段１３’により同様に各波長で最適な検出開口数で検出される。このようにして各検出波長帯の検出光は色補正レンズ１０８および１０８’により色ずれのない像を一次元又は二次元撮像装置１４および１４’の撮像面に結ぶ。
【００３３】
撮像装置１４、１４’で検出されたアライメントパタ−ンは信号処理回路２に送られ、パタ−ンの位置が検出される。このようにしてウエハ上のアライメントパタ−ンの位置が検出されれば、この情報と各パタ−ンを検出したときのウエハテ−ブルの位置（図示されていないが通常レ−ザ測長器により計測される）の情報から正しく重ね露光されるようにウエハステ−ジを微動調整し、重ね露光が実行される。上記の分光強度および指向性調整照明手段１２および分光透過率および開口数調整検出手段１３は後に実施例を用いて詳細を説明するが、これら手段により調整される照明光の分光強度および指向性、並びに検出の分光透過率及び開口数は検出する対象のウエハパタ−ンの構造や材質特性により一般に最適値が異なる。これら最適値は予めオフラインで計算し、その値を入力端末２２から入力しても良いし、図１に示すように構造や材質の特性値を端末２２から入力して最適値演算回路２１で最適値を演算しても良い。
【００３４】
なお、図１の実施例では分光強度および指向性調整照明手段１２と分光透過率および開口数調整検出手段１３とを同時に使用しているが、１２における分光強度の調整と１３における分光透過率の調整はいずれか一方を行えば十分である。
【００３５】
図４は本発明のパタ−ン検出方法をアライメント評価装置に適用したものである。図１と同一部品番号は同一物を表わす。ウエハテ−ブル７の上のウエハには既に重ね露光、現像された例えばレジストパタ−ン４１２が図５のように形成されている。即ち、４１２は重ね露光することにより形成されたパタ−ンである。他方パタ−ン４１１は重ね露光される前に形成されていたパタ−ンである。ｘ方向の重ね合わせ精度を評価するには、図５のＸＸ’断面を示した図６の左右のパタ−ン４１１の検出中心Ｃｐと左右のパタ−ン４１２の検出中心Ｃｒのずれを求めれば良い。しかしながら図６に示したように通常のパタ−ンは多層構造になっており特に最上層が光学的に透明な層であると、多重干渉の影響を受け最適な波長分光特性を持った光で照明しないと、検出波形が歪み誤検出する。この実施例ではパタ−ン４１１とパタ−ン４１２は異なる層構造であるから、両者に対し総合的に最適となる波長を選ぶ。
【００３６】
図４で１１’は照明系であり、ここを出射した光は分光強度および指向性調整照明手段１２で所望の分光強度と、各波長に応じた指向性を持っている。即ち後に説明するように、波長ごとに大きさと強度が異なる２次光源を分光強度および指向性調整照明手段１２で作る。このような光がレンズ１０１’を通り、ビ−ムスプリッタで反射後、対物レンズの瞳にこの２次光源の像を結ぶ。対物レンズの瞳を通過した光はウエハ上のパタ−ンをテレセントリックに照明する。ウエハから反射した光は対物レンズ５’ビ−ムスプリッタ１０５’結像レンズ１０９を通り、２次元撮像装置１４’に結像される。結像レンズ１０９と２次元撮像装置１４’の間には分光透過率および開口数調整検出手段１３が配置されており、検出光の波長に応じた開口数で検出するようになっている。
【００３７】
上記実施例では分光強度および指向性調整照明手段１２および分光透過率および開口数調整検出手段１３でともに分光強度を変化させているが、一方のみで十分である。即ち例えば分光強度および指向性調整照明手段１２で照明光の分光強度を最適に調整すれば、分光透過率および開口数調整検出手段１３は各波長にたいし同一の透過率で良い。
【００３８】
本実施例で示す方法により検出された２次元パタ−ンを２次元撮像装置で検出し、ｘ方向とｙ方向のパタ−ン４１１とパタ−ン４１２のそれぞれの中心のずれを求めれば、アライメント精度が求まる。ウエハの層構造や材質に依存して最適な照明光分光強度及び検出時の光学系の開口数、或いは照明光の指向性は異なるため、入力端末２２からこれら条件を入力すると、処理回路２’で最適な照明光分光強度及び検出時の光学系の開口数、或いは照明光の指向性が計算され分光強度および指向性調整照明手段１２、および分光透過率および開口数調整検出手段１３を駆動し、そのような照明光分光強度及び検出時の光学系の開口数、或いは照明光の指向性が実現する。
【００３９】
図７は本発明のパタ−ン検出装置の分光強度および指向性調整照明手段１２の具体的実施例を示したものである。また、図７は後述するように一部の構成を変えることにより、本発明の分光透過率および開口数調整検出手段１３の一実施例にもなりえる。１２１は回折格子或いはプリズム（図示せず）等の分光手段である。検出用照明光源１１（１１’）からの出射光は図のｙ方向には指向性をある程度高めた状態の光１０１０となり、分光手段１２１に入射する。検出用照明光源は検出波長域を総て含む分光強度を持っているが、検出に最適な分光強度特性を持っていない。そこで分光手段１２１で分光されたビ−ム１０１１Ａから１０１１Ｎを集光レンズ１２３により空間光透過率変調器１２５上に各分光光を集光するように分離する。回折格子を用いた場合、１０１２Ａには最も波長の短い光が、１０１２Ｎには最も波長の長い光が集光する。なお分光手段１２１に照射された光の０次透過光は用いないので遮光板１２６により遮光する。この遮光板は図７に示すようにレンズ１２３の前に配置しても良いし、空間光透過率変調器１２５上に配置しても良い。
【００４０】
そこで空間光透過率変調器１２５を図８（Ａ）に示すように遮光部１２５２に透過部１２５１を設けた構造で、透過部１２５２は各波長の光の強度がここを通過後最適な強度になるよう、図８（Ｂ）に示すように透過率Ｔ（λ）を変化させる。このように部分的に光の強度を変化させる方法としては、例えば液晶シャッタマトリックスや、ＰＺＴ、フォトクトミック材料等がある。またもっと簡単な物として、ｙ方向の変化をステップ的に変えても本発明の目的が達成できるため、図１９（Ｂ）に示すようにｘ方向に透過率が少しずつ変化するｘ方向に長い短冊状のＮＤフィルタ−をＸ方向に移動可能な構造にし、図１９（Ａ）に示すようにｙ方向に並んだ沢山の短冊をそれぞれｘ方向に制御し、図８（Ｂ）に示すような所望の透過率になるようにしても良い。
【００４１】
次に式（１）を用いて前述したように各検出光に対し同一の解像度で検出させるための照明光の指向性を最適化する実施例について図９を用いて説明する。即ち各検出光に対し同一の解像度で検出させるには、各検出波長λに比例した検出光学系の開口数ＮＡにし、さらにこのＮＡに対し各波長での照明光のパ−シャルコヒ−レンシ−が一致するようする。図９（Ａ）に示す空間光透過率変調器１２５’の空間光透過率変調領域１２５３の内１２５２を光遮光領域１２５１を光透過領域にし、波長の長いλ_Ｂでは波長の短いλ_Ａに比べ像検出のＮＡが大きく、またこのＮＡに比例し照明のパ−シャルコヒ−レンスが大きくなるように、図９（Ａ）に示す照明用の空間光透過率変調器１２５’透過領域の幅を広くしている。この幅の比はほぼ波長の比に等しくしている。液晶シャッタマトリックスや、ＰＺＴ等を用いることにより空間光透過率変調領域１２５３の領域内を電気信号で駆動することにより各波長ごとに所望のＮＡ（後述）とこれに比例した照明のパ−シャルコヒ−レンスで検出することを可能にする。
【００４２】
空間光透過率変調器１２５、或いは１２５’を透過した光は図７の集光レンズ１２４により回折格子或いはプリズム（図示せず）等の分光手段１２２に入射する。分光手段１２２の面はレンズ１２３と１２４により分光手段１２１の面と共役な関係にあるので、分光手段１２１に入射する光、即ち上記図のｙ方向には指向性をある程度高めた状態の光１０１０の強度分布が分光手段１２２上に再現される。但し、空間光透過率変調器１２５で変調されたことによる各波長ごとに強度のの変化を受けている。その結果分光手段１２２を透過し、回折する光は所望の照明のパ−シャルコヒ−レンスと分光照度分布を持ってウエハ４を照明することになる。
【００４３】
図７では１２を照明系に用いる場合について説明したが、図７の１２と同じ構成で１３を検出系に用いることにより、上記の各波長に対し所望の検出ＮＡを与え、各波長での検出像の解像度を同一にすることを可能にする。即ちこの場合には図７の１３は例えば図１の実施例の１３及び１３’の位置に配置され、ウエハ上のパタ−ン４１で反射した光が縮小レンズ５、ミラ−１０５、ビ−ムスプリッタ１０３を通り、ビ−ムスプリッタ１０６またはミラ−１０７で反射し、図７の分光手段１２１上に結像し、各波長ごとに異なる回折角で回折した後、レンズ１２３、空間光透過率変調器１２５を通り、レンズ１２４で分光手段１２２上にパタ−ン４１の像を再び結像する。
【００４４】
空間光透過率変調器１２５は図９（Ａ）に示すように光透過領域１２５１の幅が各波長ごとに、即ち高さ方向ｙに沿って式（１）を満たすように変化している。但し照明系内で図８（Ｂ）や図９（Ｂ）に示すように各波長ごとに透過率を変化させている場合には、検出系内の空間光透過率変調器１２５’は波長ごとの透過率変化は必ずしも必要でない。上記レンズ１２４で分光手段１２２上に再結像されたパタ−ンは各波長で異なる入射角で分光手段１２２に入射、結像するがここを透過する回折光１０２０は各波長の光が同一の光路を進む。この回折透過光は図１の色補正レンズ１０８または１０８’により一次元又は二次元撮像装置１４及び１４’の撮像面に分光手段１２２上の像を再結像する。
【００４５】
色補正レンズ１０８または１０８’はそれぞれの波長バンド幅内の光にたいし色収差を含む各種収差が補正されており、それぞれのバンド幅の光の像が一次元又は二次元撮像装置１４及び１４’の撮像面に歪及び色ずれなく結像する。
【００４６】
次に、上記に説明したパタ−ン検出装置の実施例の構成を用いて、具体的にパタ−ン検出する方法を示す。パタ−ン検出する対象は図３に示すような層構造を持っているため、このパタ−ンの最上層がフォトレジストの場合、フォトレジストの厚さは図１０に示すようになっている。
【００４７】
フォトレジストは通常スピンコ−タ−で塗布されるため、図３の左上にウエ−ハの中心があればフォトレジストの流れは左から右に向かう。その結果フォトレジストの厚さは図１０（Ａ）に示すように、パタ−ンエッジの左端に比べ右端の方が厚くなる。この結果フォトレジストの厚さの範囲は図１０（Ａ）に示すようにｔ_Ｂ１からｔ_Ｂ２までの厚さと、ｔ_Ｔ１からｔ_Ｔ２までの厚さの範囲が存在することになる。
【００４８】
このようなレジスト構造、層構造、およびレジストや下地のパタ−ンの材質の光学的定数（複素屈折率等）が分かれば、図１２に示すようにこれら条件をパラメ−タにして干渉計算を行う。即ち上記のパラメ−タに対して波長を変えたときに、干渉強度がどのようになるか計算する。得られた結果から、如何なる分光照明強度分布で照明すればず１０（Ａ）のレジスト厚の範囲、即ちｔ_Ｂ１からｔ_Ｂ２までと、ｔ_Ｔ１からｔ_Ｔ２までで干渉強度の変化が小さく、一定値に近づくかを計算により求めることにより、分光照明強度を決定する。このように決定された分光照明強度の分布Ｉ_ｉ（λ）は例えば図１１の（Ａ）に示すようになっている。この分布はあくまでもウエハを照明する光の分光照度分布であり、実際には光源の分光強度分布や、光学系の分光透過率分布は一定でないため、その補正をする必要がある。
【００４９】
即ち例えば光源の分光強度分布Ｉ_ｉｉｌ（λ）が図１１（Ｂ）のようになっており、光学系の分光透過率分布Ｉ_ＯＳ（λ）が図１１（Ｃ）のようになっていれば、空間光透過率変調器１２５の各波長ごとの透過率Ｉ（λ）は図１１（Ｄ）に示すように次式で求まる値にする必要がある。
【００５０】
【数２】

【００５１】
このようにして決定されたＩ（λ）になるように空間光透過率変調器１２５を駆動すれば図１０（Ｂ）に示すようにレジスト厚の範囲、即ちｔ_Ｂ１からｔ_Ｂ２までと、ｔ_Ｔ１からｔ_Ｔ２までで干渉強度の変化が小さくなり、検出パタ−ンの波形はレジストの塗布むらの影響を受けず、対称になり、精度の高いパタ−ン位置検出が実現する。
【００５２】
本発明のパタ−ン検出法をパタ−ン位置検出装置に実施した例を図１６を用いて説明する。本実施例のパタ−ン位置検出装置はステッパのオフアクシスアライメント検出系、あるいはアライメント評価用の検出系に適用される。ウエハ４の上にある検出マ−ク４１は図１６（Ｂ）に示すような構造になっている。検出用の対物レンズ５’には照明光１１００がビ−ムスプリッタ５１、結像レンズ５１および対物レンズ５’を通して照明される。反射光は照明の光路を逆に戻りビ−ムスプリッタで反射されて検出光１２００となる。検出光は分光手段１２１の上に検出パタ−ンの像を結ぶ。分光手段をまっすぐ通過する光は遮光版１２６で遮光される。
【００５３】
一方分光手段で回折された光は波長ごとに回折角度が変わり集光レンズ１２３上では短い波長の光はＤ_{λ １}に、長い波長の光はＤ_{λ ｎ}に集光する。集光レンズ１２３を透過した各波長の光は互いに平行になり、開口数決定開口１１５１を通過する。これにより前述のとおり波長に応じてＮＡが変わるようになる。開口数決定開口１１５１通過した光はシリンドリカルレンズ１２４’に入射する。この非回転対称結像系であるシリンドリカルレンズ１２４’はパタ−ン位置検出方向ｘに対しては結像作用を持ち、それと直角の方向は結像作用を持たない。従って、撮像装置１４”の撮像面１４０上にはｙ方向は分光されたままの状態の分光一次元像、即ち、ｘ方向はウエハ上のパタ−ンが拡大された結像状態で像が形成され、この像が各色ごとに縦方向ｙに並んで結像されている。
【００５４】
即ち波長λ_１からλ_ｎまでの各波長の光による像が撮像面上に１４０上Ｐ_{λ １}からＰ_{λ ｎ}の位置に形成されている。従って、信号処理装置２１のテレビモニタ−２１０上には各波長に応じてＰ_{λ １}’……、Ｐ_{λ ｉ}’……、Ｐ_{λ ｎ}’の位置に各波長の像が形成され、各波長での検出信号が得られている。
【００５５】
このように各波長ごとに分離検出された信号を用いて信号処理回路２１でパタ−ン検出する方法を以下に示す。撮像装置１４”で検出された各波長の検出波形のうちλ_１、λ_ｉ、およびλ_ｎの各波長での検出波形は例えば図１７（Ａ）に示すＩ_{λ １}（ｘ）、Ｉ_{λ ｉ}（ｘ）、Ｉ_{λ ｎ}（ｘ）のようになっている。これらの検出波形のレベルは光源の分光強度分布Ｉ_ｉｌｌ（λ_ｉ）や、光学系の分光透過率分布Ｉ_ＯＳ（λ_ｉ）の影響を受けている。更にパタ−ン各部のレジスト膜厚や下地の構造、材質から前述したように最適な分光照明強度Ｉ_ｉ（λ_ｉ）になるようにするには上記の各波長の検出波形Ｉ_{λ １}（ｘ）、Ｉ_{λ ｉ}（ｘ）、Ｉ_{λ ｎ}（ｘ）に次に示す補正値α（λ_ｉ）を掛けたものを最適な検出波形とすればよい。
【００５６】
【数３】

【００５７】
従って、最終的に得られる最適な検出波形Ｉ（ｘ）は次式で求まる。
【００５８】
【数４】

【００５９】
このようにして求まった検出波形は図１７（Ｃ）に示すようになる。
【００６０】
次に本発明の信号処理に関する他の実施例を説明する。図１６の検出光学系は検出光に対し、色収差補正された光学系を用いているが、通常の顕微鏡対物レンズでは目視観察用に用いるため、対象物を目視観察したときに色のにじみが出なければ十分仕様に耐える。しかしアライメントの評価ではアライメントの精度を１０ｎｍ程度の精度で評価する必要がある。そのため目視では分からない程度の色収差でも評価精度に影響を与える。そこで図１６の光学系が組立られた状態で色収差がどの程度残っているかを評価し、アライメントの精度評価のときに、この残存色収差を補正し、色収差が０の状態で検出評価することが重要になる。
【００６１】
本実施例ではこれを実現している。即ち図１６の光学系が組立られた状態で、照明光源、又は撮像装置の撮像面の前に干渉フィルタ−等を入れ単波長で検出したときの色収差をを評価しておく。例えば、Ｓｉウエハに決まった寸法の凹パタ−ンを形成しておき、これをいろいろな単波長、またはバンド幅の狭い光で検出し、検出位置とパタ−ン幅或いはパタ−ン間隔を計測しておく。図１８（Ａ）はこのようにして色収差を評価した結果の例である。即ち横軸は波長、縦軸は横倍率を表わしている。このように波長λ_１、λ_ｉ、およびλ_ｎ等の波長で横倍率が異なっていると、図１６の実施例で検出される各波長での検出信号は，図１８（Ｂ）に示すようにパタ−ン幅がＷ_{λ １}、Ｗ_{λ ｉ}、およびＷ_{λ ｎ}のように異なってくる。
【００６２】
しかしこれら各波長での波形は独立に得られているから、上記方法で求められている横色倍率Ｍ（λ_１）、Ｍ（λ_ｉ）、Ｍ（λ_ｎ）等を用いて倍率の補正を行うことができる。図１８（Ｃ）はこの補正の結果を表わしている。即ち補正された結果はどの波長においてもパタ−ンの幅もしくは間隔は等しく、検出位置も一致させるようにすることが可能になる。
【００６３】
このように倍率の補正を行った後、図１７で説明した分光照明強度等の補正を行えば、図１８（Ｄ）に示すように理想的なパタ−ン検出が実現する。即ち本実施例では光学系では完全に取り除くことが不可能な色収差を、分光検出することにより各波長での波形を分離検出し、各波長の波形に対し、検出位置座標対応の補正（色倍率補正、検出中心の色ごとのオフセット（位置ずれ）、色ごとの像歪等に起因する色ごとの位置ずれの補正）を行い、更に各検出信号毎にレベル補正を行うことが可能になった。
【００６４】
また図１６のパタ−ン検出装置は上記の補正のほかに、例えば照明光の波長ごとに照明むらが発生しているような場合でも、各波長ごとに検出波形が得られるため、各波長ごとにこの照明むらを補正することが容易にできる。
【００６５】
更に図１６のパタ−ン検出装置の優れている点を説明する。図１６（Ｂ）や図１０に示すように凹パタ−ンの周辺とパタ−ン部（凹んだ部分）では図１０（Ａ）に示すようにレジストの厚さが異なっている。周辺部のレジスト厚に対し最適な照明の分光強度があり、これはパタ−ン部のそれとは必ずしも一致しない。このような場合、本実施例では各波長ごとに検出波形が得られているため、各波長のそれぞれの部分で異なる上記（３）式のα（λ_ｉ）を用いる。即ちα（λ_ｉ）を座標ｘの関数として扱えば、レジストの場所による厚さの違いに応じてα（λ_ｉ）を最適な値にでき、歪みのない検出波形を得ることができるようになり、最も高い精度で検出することが可能となる。なおこのようにα（λ_ｉ）を座標ｘの関数α（λ_ｉ，ｘ）にする場合、例えば図１０（Ａ）に示すようにパタ−ンエッジ部ｘ_Ｌ、ｘ_Ｒにおいて不連続になる。このような不連続部が生じると検出波形に不連続部や歪みが発生するため、不連続部でスム−ズに連続化する処理を行う。
【００６６】
図１６の実施例ではｘ方向のパタ−ンを検出する光学系しか表示されていないが、ｘｙ２軸を検出する場合には分光手段１２１と結像レンズ５１の間にビ−ムスプリッタを挿入し、ｘ方向と同様の（ただしｙ方向を検出する構成である）検出系を設ければよい。
【００６７】
上記の図７で本発明のパタ−ン検出装置の分光強度および指向性調整照明手段１２と分光透過率および開口数調整検出手段１３を説明し、また図１６では分光透過率および開口数調整検出手段１３を説明したが、分光透過率および開口数調整検出手段で設定された各波長に対する最適な分光指向性が存在する。即ち、上記したようにどの検出波長に対しても同一の解像度を持たせるため（１）式に示した解像度Ｌが検出する総ての波長にたいしほぼ同一の値になるように、ＮＡを波長ごとに変え、このＮＡ（λ）に対して決まる照明のパ−シャルコヒ−レンスσ（被検物体が平坦な時に、検出の開口数を決定する開口、即ち、１２５１の上での被検物体の反射光の広がりとこの開口の幅の比）が波長によらずほぼ一定にする。このようにすれば、どの波長に対しても、ＭＴＦ（Ｍｏｄｕｒａｔｉｏｎ Ｔｒａｎｓｆｅｒ Ｆｕｎｃｔｉｏｎ）が同じになり、歪みの少ない波長合成波形が得られることになる。
【００６８】
次に本発明で２次元的なパタ−ンをｘｙ分離してそれぞれの方向を独立に検出するのではなく、２次元的にｘｙ同時に検出する実施例を図２０を用いて示す。
【００６９】
図２０は２次元同時検出に用いる照明系の実施例である。検出用光源１１’は水銀ランプ或いはキセノンランプ１１１から出射する光を楕円面鏡１１２でミラ−１１３を通過させ光ファイバ−１２１の入射端に集光させる。光ファイバ−１２１はその途中で複数のファイバ−束に分岐している。複数の分岐ファイバ−の光出射端１２１１、１２１２、１２１３、１２１４、１２１５は図示するように光軸方向の位置がずれるようにしている。このファイバ−の光出射端の直後に波長選択フィルタ−および指向性調整照明手段１２２１、１２２２、１２２３、１２２４、１２２５が設置されている。
【００７０】
即ち１２２１は波長が６４０ｎｍから７００ｎｍの光を通すとともにＮＤフィルタ等を重ねて用いることにより所望の光強度にする。波長選択フィルタ−および指向性調整照明手段１２２１等は図２０（Ｂ）に示すように遮光部１２２０の内側に実線で示した開口１２２１Ａが開いている。
【００７１】
同様に１２１２は５８０ｎｍから６４０ｎｍの光を通す。以下１２１３は５２０ｎｍから５８０ｎｍ、１２１４は４６０ｎｍから５２０ｎｍ、１２１５は４００ｎｍから４６０ｎｍの光を通すとともにＮＤフィルタ或いは電気信号で透過率を変化させることができる液晶シャッタ等を重ねて用いることにより所望の光強度にする。また例えば１２１５では図２０（Ｂ）の点線で示した開口１２２５Ａが開いている。この各波長選択フィルタ−および指向性調整照明手段を通過した光は波長選択ビ−ムスプリッタ１２３に入射する。
【００７２】
反射面１２３１は全反射であるが、例えば１２３３面は５８０ｎｍ以下の波長の光を反射し、５８０ｎｍ以上の波長の光を透過する。このような波長選択ビ−ムスプリッタ１２３を通ってきた光を検出光学系の照明光に用いる。例えばこの照明系１２’を図４の実施例に用いる場合を例に採ると、開口１２２１Ａ〜１２２５Ａがレンズ１０１’によりパタ−ン検出レンズ５’の入射瞳に結像され、この像の径に比例した照明指向性を持ったケ−ラ照明光を被検出パタ−ンに照射できる。
【００７３】
このようにして各波長帯の光に分離し、各波長帯の光を所望の強度比にし、かつ各波長帯の光を所望の照明指向性にして被検物に照射することが可能になる。本実施例では照明の指向性が光軸に回転対称性を有しているため、被検物のパタ−ンが一次元的であっても二次元的であっても、関係無く、望ましい照明を実現することが可能である。
【００７４】
次に２次元同時検出に用いる検出光学系の実施例を図２１に示す。
【００７５】
図２０の実施例の分光強度および指向性調整手段１２’を用いて照明されたウエハのパタ−ンからの反射光は図４等で示されている検出用対物レンズを通り、拡大結像される。この像を２次元撮像装置１４’で検出する前に図２１に示した分光透過率および開口数調整検出手段１３”を通す。この分光透過率および開口数調整検出手段１３”を図４の実施例に適用する場合には結像レンズ１０９のほぼ後側焦点位置に上記の各波長帯の分光透過率および開口数調整検出手段１３３１、１３３２、１３３３、１３３４、および１３３５が来るように配置する。ここに至るまでに、ビ−ムスプリッタ１３１により各波長を分離する。このビ−ムスプリッタは基本的に上記図２０の１２３と同じである。
【００７６】
上記の各波長帯の分光透過率および開口数調整検出手段１３３１、１３３２、１３３３、１３３４、および１３３５には図２１（Ｂ）に示すように遮光部１３３０の中央部に開口１３３１Ａ（〜１３３５Ａ）が開いており、この開口径で検出の開口数ＮＡが決まる。ここを透過した各光は再びビ−ムスプリッタ１３２で合成され、二次元撮像装置１４’に像を結ぶ。なおビ−ムスプリッタ１３２はビ−ムスプリッタ１３１や１２３とは異なり、例えば１３２３面は５２０ｎｍ以下の波長の光を透過し、５２０ｎｍ以上の波長の光を反射する。開口１３３１Ａ（〜１３３５Ａ）は各波長帯での解像度がほぼ等しくなるように、例えば数式１に示される解像度Ｌが各波長帯の中心でほぼ等しくなるように、その径が決定されている。このようにすることにより各波長でほぼ同一の解像度で、また被検出物の構造、材質に応じて最適な分光照明特性でパタ−ンを検出することが可能になる。
【００７７】
図２２（Ａ）は本発明の実施例で、図２１に示した実施例とは異なり各波長帯に分離し、被検出パタ−ンから反射した検出光を各波長帯に分離したまま別々に二次元撮像装置で検出するものである。ビ−ムスプリッタ１３１で各波長帯に分離された光は開口数調整検出手段１３３１’、１３３２’、１３３３’、１３３４’、および１３３５’で各波長帯ごとに所望の検出開口数ＮＡされた後結像レンズ１０９１、１０９２、１０９３、１０９４、および１０９５により二次元撮像装置１４１’、１４２’１４３’、１４４’、及び１４５’上に結像される。各二次元撮像装置で得られた各波長帯の検出信号は信号処理回路２’に入力される。
【００７８】
図２２（Ｂ）に示す各撮像装置の撮像面上の任意の点Ｐの座標（ｘ’，ｙ’）は検出光学系の持っている色収差等の収差の影響を受け、検出光学系が無収差の理想的な光学系の場合の結像点Ｐ_０（ｘ，ｙ）とは異なっている。いまｉ番目の波長帯をλ_ｉで表し、波長帯λ_ｉの結像系での上記点Ｐ（ｘ’，ｙ’）と点Ｐ_０（ｘ，ｙ）の関係が数式５で表せるとする。
【００７９】
【数５】

【００８０】
ここで（ｘ，ｙ）（ｘ’，ｙ’）は二次元座標上の位置ベクトルであり、ｐ_{λ ｉ}は波長帯λ_ｉごとに異なる座標変換の関数、即ち収差による歪を表す関数である。関数ｐ_{λ ｉ}（ｘ’，ｙ’）は例えばウエハ上に規則正しい碁盤目のパタ−ンを形成し、このパタ−ンを用いて検出し、検出像の位置信号から求めることができる。
【００８１】
図２３は二次元パタ−ンを検出する他の実施例である。図２０で示した照明光源を用い（図示せず）、ファイバ１２１’の出射端は図１０の実施例のように分岐せず、照明光を出射させる。出射光は図２３（Ｂ）に示す指向性調整手段１２２’を通り、波長選択手段１２５５を通過する。指向性調整手段１２２’は前述した照明の指向性を各波長ごとに所望の値にする。このため図２３（Ｂ）に示すように液晶等の透過率可変領域１２２０’の内所望の領域１２２１Ａ’（波長がλ_ｉの場合）の内側を所望の透過率にし、外側を透過率０にする。
【００８２】
波長選択手段１２５５には図２３（Ｄ）に示すように透過波長帯が少しずつ異なっているバンドパスフィルタ１２５５１〜１２５５８が円盤上に円周方向に配列されている。例えば４００ｎｍ〜７００ｎｍを８等分した波長帯λ_１〜λ_８の光が円盤を回転することにより得られる。このようにして得られた照明光を上記の実施例と同じようにウエハ等の被検物に照射し、得られた反射光を上述の１２２’と同じ機能を持つ開口数調整検出手段１３３’に導く。１３３’は検出光学系の瞳と共役な位置にあり、光透過領域１３３０’中の１３３１Ａ’内を透過状態、外を遮光状態にすることにより、所望の検出開口数ＮＡを得ることができる。
【００８３】
波長帯λ_１での検出が完了し検出情報Ｉ_{λ ｉ}（ｘ’，ｙ’）を信号処理回路２”に記憶したならば、波長選択手段１２５５の円盤を回転し、波長帯λ_２の検出ができるようにし、指向性調整手段１２２’および開口数調整検出手段１３３’を駆動し、この波長帯での決められた条件に設定し、検出、記憶する。以上の動作を順次波長帯λ_８まで行い、総べての波長帯での検出信号の記憶が終了する。得られた信号を如何に用いるかの具体的な実施例を次に説明する。
【００８４】
図２０から図２３を用いて説明した方法により得られた例えば前述の図５に示す２次元パタ−ン４１’の４１１と４１２のｘ方向およびｙ方向のずれΔｘおよびΔｙを求める方法を図２４を用いて説明する。図２４は検出信号の処理の流れを表すと共に、処理の内容を表している。各ブロックは総べて個別に専用の電子回路で実現することもできるが、処理時間が短い部分は共通のマイコンで処理する。波長帯λ_ｉの検出で得られた検出信号をＩ_{λ ｉ}（ｘ’，ｙ’）とする。パタ−ン４１１と４１２は前述したように構造及び材質が異なっているため、検出の最適分光照度分布は異なる。
【００８５】
実施例図２０、２２、２３では各波長帯で独立に検出波形が得られるので、上記最適分光照度分布に成るように、得られている信号にパタ−ン４１１と４１２ごとに、及び検出波長帯λ_ｉごとに定数α_{１ λ ｉ}、α_{２ λ ｉ}を掛けるとともに、前述の収差の補正を２２１および２２１’で行う。得られた補正後の波形は図２４の２２１および２２１’にそれぞれＩ_{１ λ ｉ}’（ｘ，ｙ）およびＩ_{２ λ ｉ}’（ｘ，ｙ）の式でで示されている。各波長で得られた補正後の波形を２２２および２２２’で総べての波長にわたりパタ−ン４１１と４１２に対し、それぞれ位置座標ごとに強度波形を足し合わせ理想的なＩ_１（ｘ，ｙ）およびＩ_２（ｘ，ｙ）を得る。
【００８６】
この得られた波形を用いて例えば対称性パタ−ンマッチングのアルゴリズムを用いてパタ−ン４１１および４１２のそれぞれのｘおよびｙ方向の中心位置Ｃ_１ｘ、Ｃ_１ｙ、Ｃ_２ｘおよびＣ_２ｙを２２３および２２３’で求める。図２４でｆ_ｘおよびｆ_ｙはパタ−ン中心を求める演算関数（例えば対称性パタ−ンマッチングの演算関数）を表す。このようにして得られたパタ−ン４１１と４１２のｘ及びｙ方向の中心から、両パタ−ンのｘ、ｙ方向の位置ずれΔｘ、Δｙは２２４で求められる。次にこのようにして得られる位置ずれ情報を如何に用いるかについての実施例を示す。
【００８７】
図２５は本発明の露光システムを示す実施例図である。１は露光装置であり、例えば図１で示される縮小露光装置である。２はこの露光装置を制御する制御回路である。１０は例えば図４で示したパタ−ン検出装置或いはアライメント評価装置である。２’はこの装置の制御回路である。９１、９２、９３はそれぞれレジスト塗布装置、現像装置、およびエッチング装置である。８はこれらフォトリソグラフィシステムを制御する制御回路である。図中点線で示した流れはウエハの流れ、実線で示した流れは信号の流れである。
【００８８】
レジスト塗布装置９１でレジストを塗布されたウエハは露光装置１に搬入される。搬入されたウエハには露光チップ毎に図２に示すようにアライメントパタ−ンが記録されているので、このアライメントパタ−ン４１（４１’）を縮小レンズ５を通して検出系１０Ｘ（１０Ｙ）で検出する。このようにして検出した位置情報とウエハステ−ジのレ−ザ−測長機（図１に図示せず）の位置情報ならびにレチクル３の位置情報を基にウエハステ−ジ７を駆動し、アライメントを行った後露光を行う。
【００８９】
このような露光をウエハ全面にわたりステップアンドリピ−トで行い、ウエハ全面にレチクル３に描画された回路パタ−ンを露光する。図１及び検出系１０Ｘの実施例に付いて既に説明したように、本発明のパタ−ン検出方法を搭載している露光装置では、正しいアライメントパタ−ン検出が行われ、正確な重ね露光が実現されている。露光されたウエハは現像装置９２に送られ、現像される。現像されたウエハには予め重ね露光の評価をするためにウエハ上に例えば図５の４１１のパタ−ンが露光前に記録されている。このパタ−ン４１１に新たなレジストパタ−ン４１２が露光されているため、これら４１１と４１２のパタ−ンの相対的な位置を１０のアライメント評価装置を用いて前述の実施例で説明したように検出することにより、重ね露光の精度が評価される。評価結果は制御回路２’から出力され、この出力２３は制御回路８に送られる。
【００９０】
制御回路８はレジスト塗布機９１及び露光装置１につながっており、重ね露光の評価結果が目標の精度に入っていなければ、その原因を分析する。その原因がレジスト塗布機にある場合にはレジスト塗布条件、例えばスピンコ−トの回転速度、回転の加減速条件、塗布機のレジスト溶媒蒸気圧或いは温度等の条件を最適になるように制御する。また重ね評価結果の悪い原因が露光装置にあるならば、露光装置内のどこに原因があるかを判断する。
【００９１】
もともと露光装置内に異常チェック機能や、アライメント検出結果や露光時の位置情報等が記録され、装置の状態が常時モニタ−、記録されている。これらの情報を制御回路８に送り、この情報とアライメント評価装置１０から送られてきた情報を用いて、露光装置１のアライメント検出系に原因があるか否かを判断する。この判断は学習機能のある制御回路８で大方の判断が可能である。本実施例のように露光装置のアライメント検出系の中に前述したような照明手段及び検出手段を具備していれば、かなりの精度の重ね露光が実現されているはずであるが、さらに高い精度の重ね露光が要求されていたり、露光装置に前述したような照明手段及び検出手段を具備していない場合には、アライメント評価結果を基に露光装置にフィ−ドバックし、以降の露光では高いアライメント精度が得られるようにする必要がある。このフィ−ドバックの方法としては２つの方法が有効である。
【００９２】
第一の方法はアライメント評価の結果得られたｘ、ｙ及びθ（θは露光チップのｒ離れた２か所にある２個のｘ方向検出パタ−ンの位置ｘ_１、ｘ_２を検出し次式でもとまる、θ＝（ｘ_１−ｘ_２）／ｒ）方向の重ね誤差の値Δｘ、Δｙ、Δθを次回の露光時にオフセット値として与え、ソフト的に補正を掛ける。即ち、このウエハを露光したときのオフセット値が例えばΔｘ_０、Δｙ_０、Δθ_０であったとすると、このフィ−ドバックを掛けて次回に露光するときのオフセット値Δｘ_１、Δｙ_１、Δθ_１をΔｘ_１＝Δｘ_０−Δｘ、Δｙ_１＝Δｙ_０−Δｙ、Δθ_１＝Δθ_０−Δθにして露光する。このようにすることによりフィ−ドバック後の重ね合わせ精度は大幅に向上する。
【００９３】
第二の方法は露光装置のアライメント検出系の前述の照明系の分光照度分布及び照明の指向性、並びに検出光学系の各波長ごとの検出の開口数等の条件をさらに最適になるように変える。以上のいずれかの方法を用いてアライメント評価結果を露光装置、或いはレジスト塗布装置にフィ−ドバックすることにより、重ね合わせ露光精度は大幅に向上する。
【００９４】
次に本発明の露光システムの実施例を図２６を用いて説明する。本図と図２５が同一番号は同一物を表している。図中の点線の矢印はウエハの流れを表しているが、ほぼ総べてのケ−スに付いて併記しているため、後の説明を判り易くするため各流れにＦで始まる番号を付けている。実線の矢印は信号情報の流れを表している。以下にこの図を用いて複数の実施例（Ａ）〜（Ｅ）について説明する。
【００９５】
（実施例Ａ）
ウエハの流れ：Ｆ１１→Ｆ１２→Ｆ２１→Ｆ２２→Ｆ３→Ｆ４→Ｆ５
前の露光、現像エッチング工程で形成されたウエハアライメントパタ−ン（レジスト未塗布）をアライメント評価装置１０で評価し、ウエハパタ−ンの対称性等を評価する。通常は対称性が良いので、この場合は以降のレジストが塗布された状態で再びアライメント評価装置１０で評価する。このアライメントパタ−ンを検出した結果が非対称であれば、レジストの塗布が非対称に成されていると判断できる。
【００９６】
即ち、この場合は、前述したようにレジスト塗布の非対称に対して検出波形が対称になるように、前述の露光装置１のアライメント検出系の分光照度および指向性調整照明手段１２及び分光透過率および開口数調整検出手段１３を用いて、最適な条件で検出する様に露光装置１にフィ−ドフォワ−ドする。このようにすれば露光装置１で露光に先だって行われるアライメント検出はレジスト塗布むらの影響を受けること無く、正確な重ね合わせ露光が実現される。
【００９７】
この場合には露光現像後に再びアライメント評価装置を用いて、アライメント結果を評価する必要はない。しかし確認の意味でＦ４の代わりにアライメント結果の評価Ｆ４１→Ｆ４２を行っても良い。前述のＦ１１の後でアライメント評価装置１０で評価した結果ウエハパタ−ンの対称性が悪い場合はその結果を制御装置８に送り、エッチング装置にフィ−ドバックする（図示せず）か、この非対称性を評価しアライメントのオフセット値を決定し露光装置１にフィ−ドフォワ−ドする。
【００９８】
（実施例Ａ’）
ウエハの流れ：Ｆ１１→Ｆ１２→Ｆ２１→Ｆ２２→Ｆ３→Ｆ４→Ｆ５
ウエハの流れは実施例Ａと同一である。
【００９９】
本実施例では露光装置１のアライメント検出系には上記の本発明の手段が施されておらず、例えば水銀ランプやレ−ザ光でえられる単色光を照明光源として検出しているとする。このような場合でも本実施例を用いれば、高いアライメント精度で露光することが可能である。
【０１００】
以下にこれを実現する方法を示す。
【０１０１】
先ず、アライメント評価装置１０検出系の条件を露光装置のアライメント検出系と同一の条件となる様にする。これはアライメント評価装置１０には分光照度、照明の指向性、検出の開口数がすべてコントロ−ルできるようになっているため、露光装置のアライメント検出系と同一条件にすることは容易に実現できる。このような状態（露光装置のアライメント検出系と同一条件）でウエハのアライメントパタ−ンの位置検出を行う。
【０１０２】
この結果を例えば（ｘ_１，ｙ_１）次にウエハの位置は固定にしたまま、前述の検出の最適条件に成る様にアライメント評価装置の光学系を変え、ウエハのアライメントパタ−ンの位置検出を行う。この結果を（ｘ_０，ｙ_０）とする。
【０１０３】
これら検出結果を用いてオフセット値Δｘ＝ｘ_０−ｘ_１，Δｙ＝ｙ_０−ｙ_１を求め、この値を次に行う露光の際のアライメント検出結果に加えることにより、高い精度で重ね露光を行うことができる。露光後のフロ−は実施例Ａと同一である。
【０１０４】
（実施例Ｂ）
ウエハの流れ：Ｆ１→Ｆ２１→Ｆ１２→Ｆ２１→Ｆ２２→Ｆ３→Ｆ４→Ｆ５
ウエハ上に形成されているパタ−ンの対称性が安定に保たれている場合には、レジストを塗布する前にアライメントパタ−ンを評価する必要がない。この場合はレジスト塗布後の評価をアライメント評価装置で行い、前述したように露光装置のアライメント検出系の照明、検出条件を最適化する。以降のフロ−は実施例Ａと同じである。
【０１０５】
（実施例Ｂ’）
ウエハの流れ：Ｆ１→Ｆ２１→Ｆ１２→Ｆ２１→Ｆ２２→Ｆ３→Ｆ４→Ｆ５
ウエハの流れは実施例Ｂと同一である。
【０１０６】
本実施例では実施例Ａ’同様に、露光装置１のアライメント検出系には上記の本発明の手段が施されておらず、例えば水銀ランプやレ−ザ光でえられる単色光を照明光源として検出している。実施例Ａ’同様にアライメント評価装置１０で露光装置のアライメント検出系と同一の検出条件で検出した結果と、最適条件で検出した結果から露光装置のアライメント検出のオフセットを求め重ね露光を行う。
【０１０７】
（実施例Ｃ）
ウエハの流れ：Ｆ１→Ｆ２→Ｆ３→Ｆ４１→Ｆ３２→Ｆ４→Ｆ５
本実施例は図２５と同一であるので、説明を省略する。
【０１０８】
（実施例Ｄ）
ウエハの流れ：Ｆ１→Ｆ２→Ｆ３→Ｆ４１→Ｆ４２→Ｆ５１→Ｆ５２
露光後に現像し、図５に示すように露光前に形成されていたパタ−ン４１１と露光により形成されたレジストパタ−ン４１２の合わせ精度をアライメント評価装置１０で評価する。評価後エッチングを行い、４１２のレジストマスクで形成されたエッチングパタ−ンと４１１のパタ−ンの合わせ精度を評価する。これら２回の評価結果が一致していれば、問題にならないが、一致していなければこの不一致量を次のウエハの露光時にアライメントのオフセット値として加えることにより、エッチングに伴い発生したアライメントオフセットを補正することが可能になり、次のウエハのパタ−ンの重ね合わせ露光を高精度に実現する。
【０１０９】
（実施例Ｅ）
ウエハの流れ：Ｆ１→Ｆ２→Ｆ３１→Ｆ３２→Ｆ４１→Ｆ４２→Ｆ５
露光後で現像する前のウエハをアライメント評価装置で評価する。この評価は露光後にレジストパタ−ンに形成されている潜像と既に形成されている下地のパタ−ンの合わせ精度を評価することになる。この評価で得られたずれ量はこのウエハを現像する前に得られるので、この結果を即座に次のウエハの露光時のアライメントオフセット値として活かすことができるため、スル−プットを落すこと無く、重ね合わせ露光を高精度に実現する。
【０１１０】
【発明の効果】
本発明により、半導体回路の露光におけるアライメントがウエハ上に塗布されたレジストの塗布むらや、下地パタ−ンの膜構造に依存せず高精度に行えるようになり、実装密度の高い回路を高い歩留で生産することが可能になった。また半導体回路の生産において実施される、形成パタ−ンの寸法、間隔等の計測が高精度に実行できるようになり、生産歩留の向上が図れる。
【図面の簡単な説明】
【図１】本発明のパタ−ン検出方法および装置及びそれを用いた投影露光装置を示すブロック図。
【図２】被検物体であるウエハ上の回路及び検出パタ−ンの平面図。
【図３】検出パタ−ンの構造を示す斜視図。
【図４】本発明のパタ−ン検出方法および装置のブロック図。
【図５】被検物体のパタ−ンの平面図。
【図６】被検物体のパタ−ンの断面図。
【図７】本発明の分光強度および指向性調整照明手段および分光透過率および開口数調整検出手段を示す断面図。
【図８】本発明の第１の空間光透過率変調器の平面図及びその透過率を示すグラフ。
【図９】本発明の第２の空間光透過率変調器の平面図及びその透過率を示すグラフ。
【図１０】（Ａ）スピンコ−タで塗布されたレジストの膜厚
（Ｂ）レジスト厚に対する多重干渉強度
（Ｃ）所望の分光照明強度の例
【図１１】（Ａ）所望の分光照明強度の例
（Ｂ）光源の分光強度
（Ｃ）光学系の分光透過率
（Ｄ）分光強度および指向性調整照明手段（または分光透過率および開口数調整検出手段）の空間光透過率変調器の透過率
【図１２】本発明における分光透過率を決定する過程を示すブロック図。
【図１３】レジスト塗むらがある場合の断面構造と従来の単波長光で検出したときの検出波形を示す図。
【図１４】下地の材質の分光反射率に波長依存性が強い場合の従来の白色光検出で検出したときの検出波形を示す図。
【図１５】本発明の作用を示す図。
【図１６】本発明のパタ−ン検出方法および装置を示すブロック図。
【図１７】検出波形を示す図。
【図１８】本発明における色収差補正を示す図。
【図１９】本発明による短冊上のＮＤフィルタにより構成した空間光透過率変調器を示す平面図及びその透過率を示すグラフ。
【図２０】二次元パタ−ンを検出する場合の本発明の照明系を示すブロック図。
【図２１】二次元パタ−ンを検出する場合の本発明の検出系を示すブロック図。
【図２２】二次元パタ−ンを検出する場合の本発明の検出系を示すブロック図。
【図２３】二次元パタ−ンの検出系を示すブロック図。
【図２４】二次元パタ−ンを検出する場合の本発明の処理フロ−を示す図。
【図２５】本発明の露光システムを示すブロック図。
【図２６】本発明の露光システムを示すブロック図。
【符号の説明】
１…露光装置、２…信号処理回路、４…ウエハ、５…縮小投影レンズ、５’…対物レンズ、８…制御装置、１０…パタ−ン検出装置、１１…光源、１４…パタ−ン検出手段、２２…信号入力端末、４１…検出パタ−ン、９１…レジスト塗布装置、９２…現像装置、９３…エッチング装置、１２１…分光手段、１２４’…シリンドリカルレンズ、１１５１…空間光透過率変調器。[0001]
[Industrial applications]
The present invention relates to a method for detecting a pattern formed on an object to be detected having a layer structure such as a semiconductor integrated circuit. In particular, the present invention relates to a method of accurately detecting information such as the position of a pattern covered with an optically transparent layer such as a photoresist and performing overexposure.
[0002]
[Prior art]
With the high integration and miniaturization of semiconductor integrated circuits, the line width of circuit patterns has become less than 0.5 μm, and the alignment accuracy in the exposure process of circuit patterns has been required to be less than 0.1 μm. Higher precision is required. In addition, the above-described accuracy or higher-accuracy detection performance is also required when evaluating the result of performing the overlay exposure after performing the alignment.
[0003]
The biggest obstacle to realizing high-precision detection is that the detection pattern itself used for exposure and detection, or a thin (1-2 μm thick) photoresist applied to this pattern does not have a symmetrical shape. That is. When the thickness of the optically transparent thin film formed on the uppermost layer of the pattern becomes asymmetrical depending on the location, the detection waveform becomes asymmetrical as shown in FIG. That is, for example, when trying to detect the center position of the pattern, the detection center is greatly shifted from the true center.
[0004]
To cope with such a problem, a method has been proposed in the past, as disclosed in JP-A-61-203640 and JP-A-1-227431, in which the detection wavelength is increased or the wavelength bandwidth is widened. I have. By adopting such a method, a considerably higher detection accuracy is obtained as compared with the case of detecting with a single wavelength.
[0005]
[Problems to be solved by the invention]
However, it is difficult for the above-mentioned conventional method to secure an accuracy enough to cope with future pattern miniaturization. This is because the detection wavelength is fixed in the above-described detection method with multiple wavelengths or a wide wavelength bandwidth, so that it is not possible to detect with high accuracy any detection target. In other words, the intensity distribution (spectral illumination intensity characteristic of light) of the detection wavelength at which the detection accuracy is highest differs depending on the layer structure of the object to be detected. In the prior art, the intensity distribution of the detection wavelength is fixed to the last, or only one or a plurality of detection light prepared in advance is used at most. This selection is made by evaluating the detection result of each light or the alignment accuracy after exposure. For this reason, an integrated circuit having a pattern line width of 0.5 μm or less cannot not only obtain sufficient accuracy but also requires a great deal of time to select a wavelength.
[0006]
This will be described in more detail. To make the story easier to understand, consider the simple structure shown in FIG. However, it is assumed that the reflectance of the base differs depending on the wavelength as shown in FIG. Such an example generally occurs in the case of a test object having a multilayer structure. Even if the underlayer does not have a layered structure, a test object having a strong wavelength dependence in the spectral reflectance characteristic of the material of the underlayer also corresponds to such an example. It is well known that when a white light having a spectral illuminance distribution as shown in FIG. 14B is used to detect a layered structure object, it is hardly affected by the film thickness, but as shown here, If the reflectance of the base varies depending on the wavelength as shown in FIG. 14 (A), the contribution to the detection waveform of a wavelength having a low reflectance becomes small, and the detection is effectively performed at a narrow wavelength as shown in FIG. 14 (C). It is no different from that. As a result, the detection waveform is more influenced by the film thickness, and the detected waveform becomes a waveform with strong asymmetry and large distortion as shown in FIG.
[0007]
Further, in the conventional detection method using multiple wavelengths or a wide wavelength bandwidth, a constant numerical aperture of the detection optical system is used regardless of each wavelength used for detection. However, if attention is paid to the fact that the synthesis of the detection waveforms obtained at each wavelength is a waveform finally obtained, the resolution L of the detection image is given by the following equation.
[0008]
(Equation 1)

[0009]
The resolution differs depending on the wavelength. As a result, the resolution changes in proportion to the wavelength, and the synthesis of waveforms at the different wavelengths is detected, thereby reducing the effect of wavelength synthesis.
[0010]
Further, conventionally, when detecting at a wavelength having a wide wavelength bandwidth, the chromatic aberration included in the detection optical system, particularly, when detecting an edge of a pattern under the influence of lateral chromatic aberration, for example, the edge position changes for each wavelength, and alignment is performed. This causes errors in detection, dimension detection, and position detection. Although such an error did not cause much problem when the detection accuracy was about 0.1 μm, it will be a serious problem in future measurement technology that requires an accuracy of 0.05 μm or less. However, it is extremely difficult to manufacture a detection optical system satisfying such accuracy, that is, a microscope objective lens and a reduction lens.
[0011]
[Means for Solving the Problems]
In order to solve the above problems, the present invention uses the following means. A light source having a wide wavelength width or a light source effectively including a plurality of single wavelengths is used as a light source, and the light emitted from this light source is subjected to a test in which at least a part of the uppermost layer on the surface of the layer structure is optically transparent. Light the object. The pattern on the surface of the test object is detected by using the light reflected by the test object. At this time, the light is emitted from the light source according to the information on the layer structure and the material of the test object. The spectral illumination intensity characteristic (wavelength intensity distribution characteristic) of the light to be emitted is effectively changed, and the pattern image of the test object illuminated at the desired spectral illumination intensity obtained by the change is one-dimensional or two-dimensional. Detect as an image.
[0012]
As a more specific means of the above-described detection method, for example, after the light emitted from the light source is changed by a spectral means so that the traveling direction of the light is changed for each wavelength, each light is separated into the separated optical path. It is sufficient to arrange a spatial light transmittance modulator that varies the light transmittance for each place where light of the wavelength passes, and to effectively change the spectral illumination intensity characteristics (wavelength intensity distribution characteristics).
[0013]
As another specific example, when a pattern image of an irradiated test object is detected as a one-dimensional or two-dimensional image, the light is reflected in the optical path of light reflected by the test object for each wavelength of the reflected light. By inserting a spectral transmittance variable filter that changes the transmittance and changing the spectral transmittance of the detection light, illumination with the optimal spectral intensity is achieved, and the spectral illumination intensity characteristics of the light emitted from the light source are obtained. (Wavelength intensity distribution characteristic) is effectively changed.
[0014]
According to the present invention, in order to solve the above-mentioned problem of the prior art in which the resolution differs for each wavelength, a pattern image of a test object irradiated with the desired spectral illumination intensity is detected as a one-dimensional or two-dimensional image. In this case, the numerical aperture (NA) of the pattern detection is made different for each wavelength of the reflected light reflected by the test object.
[0015]
Further, at this time, directivity is provided for each wavelength of the illumination light so as to achieve partial coherence of the illumination having a fixed ratio with respect to the numerical aperture for pattern detection for each wavelength.
[0016]
In order to solve the above-mentioned problem with respect to chromatic aberration, in the present invention, light emitted from a light source having a wide wavelength width or a light source that effectively includes a plurality of single wavelengths is illuminated on a test object, and light reflected by the test object is reflected. To detect the pattern on the surface of the test object. At this time, the light emitted from the light source and reflected by the object to be detected is substantially imaged on a spectral means such as a grating, and a pattern detection direction orthogonal to the spectral direction of the pattern spectrally separated by the spectral means. Forms an image of the pattern on the spectral means, and detects the spectral direction by the two-dimensional imaging means as a spectral one-dimensional image in a spectrally non-imaged state. In addition, for each signal in the pattern detection direction corresponding to each spectral wavelength of the spectral one-dimensional image, the chromatic aberration due to the detection optical system is corrected by signal processing, and then the chromatic aberration corrected signals are combined, and the combined signal is obtained. To detect the pattern.
[0017]
If the above-described method of detecting a one-dimensional spectral image is used, each signal in the pattern detection direction corresponding to each spectral wavelength is multiplied by a different constant for each wavelength to the entire signal of each wavelength, and then each wavelength is multiplied. If the signal is synthesized and the pattern is detected based on the synthesized wavelength, the spectral characteristics of the illumination light are changed, and the same pattern as that in the case where the pattern image of the test object irradiated with the desired spectral illumination intensity is detected is detected. effective.
[0018]
Further, if the above-mentioned method using a one-dimensional spectral image is used, a constant different depending on the location of the signal of each wavelength is multiplied for each signal in the pattern detection direction corresponding to each spectral wavelength. If the signals are synthesized and the pattern is detected based on the synthesized wavelength, it is determined that the optimum spectral illumination intensity is given to each location according to the layer structure and the material different for each location.
[0019]
In order to make the spectral illumination intensity desired, that is, the optimal one, the present invention uses the information on the layer structure and the material of the test object to determine the multiple interference intensity caused by the layer structure at each wavelength of the detection light. The spectral illumination intensity characteristic (wavelength intensity distribution characteristic) of the light emitted from the light source is calculated based on the information on the layer structure and material of the object to be measured so that a detection signal waveform that faithfully reflects the pattern structure is obtained. It is effectively changed according to it.
[0020]
The pattern detecting means described above is not only a means effective only for a pattern detection or a pattern position detecting device, but also a means for an alignment pattern detecting system required for alignment of a projection exposure apparatus.
[0021]
The alignment pattern formed under the resist applied to the wafer, the alignment pattern after exposure of the wafer, or the alignment pattern after development and further after etching is obtained by the pattern detection means. The exposure system is configured to accurately measure the pattern position using the method, and to feed back or feed the result to a resist coating machine, an exposure apparatus, a development apparatus, or an etching apparatus. As a result, overlay exposure is realized with higher accuracy.
[0022]
In particular, in this exposure system, a pattern detection system in the exposure apparatus, that is, a spectral illuminance (a bright line spectrum of a mercury lamp, monochromatic light such as laser light, or polychromatic light) or an illumination of an alignment pattern detection system may be used. The pattern position detection system was set under the same conditions as the optical conditions such as the directivity of the optical system or the numerical aperture of the detection optical system, and the position of the alignment pattern before exposure by the exposure apparatus was detected. The first pattern position information and the optimal detection optical system conditions for the alignment pattern being detected are detected at the same wafer position as when the first pattern position information was obtained. A difference from the obtained second pattern position information is obtained, and this difference is fed to the exposure apparatus as an alignment offset of the exposure apparatus. If the alignment detection system of the exposure apparatus is used to align the wafer, for example, Topata - without consisted in optimal conditions to down, correct offset correction is made, it is possible to realize a highly accurate superposing exposure.
[0023]
[Action]
The operation of the present invention will be described with reference to FIG. If the base has a different reflectance depending on the wavelength as shown in FIG. 14 (A), it is illuminated through a filter having a spectral transmittance as shown in FIG. 15 (A), and as shown in FIG. This is equivalent to performing perfect white illumination with a constant reflectance, that is, the condition of the multiple interference is no longer wavelength dependent, and the detection waveform is symmetric as shown in FIG. The error is reduced.
[0024]
In general, the reflectance of the underlayer is not the simple reflectance as described above. In particular, when the underlayer has a multilayer structure, the reflectance is a complex reflectance. Even in such a case, the complex reflectance can be obtained by calculation from the characteristics (spectral complex refractive index) of the underlying layer structure and material, and can be measured optically. If the optimum spectral intensity of the illumination light is determined from the complex reflectance, the thickness of the thin film on the substrate, and the refractive index, and the illumination having such spectral characteristics is realized and detected by the above-described means, highly accurate detection can be achieved. It can be performed.
[0025]
Further, in the present invention, by setting the numerical aperture of the detection optical system that differs depending on the detection wavelength to be the same for each wavelength, color shift (color bleeding) at the pattern edge portion that occurs even in the ideal optical system is completely eliminated. And complete detection is possible.
[0026]
Also, for an optical system having non-ideal chromatic aberration, the light reflected by the object to be detected is once formed on a dispersing means such as a grating, and then the separated light is formed into an image in the pattern detection direction. By forming an image on the two-dimensional imaging device in a state orthogonal to the direction perpendicular to the image, it is possible to separately detect images at each wavelength. As a result, it is possible to correct the color shift of each wavelength detected by an optical system having chromatic aberration and formed with color shift by image processing.
[0027]
Further, since the images at the respective wavelengths are separately detected, it is not necessary to make a hard correction so that the spectral illuminance characteristics of the illumination light become desired as in the above example. That is, the signals obtained by multiplying the signals of the respective wavelengths by different constants may be synthesized so as to obtain the same result as that obtained by illuminating the detected image of each wavelength with the above-described desired spectral characteristics. .
[0028]
If a method that can separately detect an image at each wavelength is used, it is possible to select the wavelength to be used for the detection light for each part according to the thickness of the thin film that differs for each part. Detection with a small detection error can be realized.
[0029]
【Example】
Hereinafter, the present invention will be described in more detail with reference to examples.
[0030]
FIG. 1 shows an embodiment in which the pattern detection method of the present invention is applied to a stepper (projection exposure apparatus). The reticle 3 is irradiated with i-line or excimer laser light from an exposure illumination system 6, and the light transmitted through the reticle is reduced by a reduction lens 5 so that a reticle drawing pattern can be moved in a xyz direction by a wafer 4 on a wafer stage 7. Is exposed on the exposure area 40 of the exposure. By step-moving the wafer stage in the x and y directions, exposed areas of the circuit pattern are arranged on the wafer 2 as shown in FIG. Alignment patterns 41 are formed between adjacent exposure circuit pattern areas in order to realize, with high accuracy, a circuit pattern that has already been formed by exposure in the next exposure process. Have been. As shown in FIG. 3, the alignment pattern 41 has a photoresist 412 applied on a base pattern having a concave portion 411. It is necessary to accurately detect the position of this pattern through the reduction lens 5.
[0031]
A detection system for detecting the position of the alignment pattern 41 in the x direction is indicated by a dotted line 10x. Although not shown in the figure, the detection system 10y for detecting the alignment pattern for the position in the y direction is in a direction perpendicular to the plane of the drawing. The detection system 10x has an illumination light source 11 for detecting a pattern, such as a halogen lamp or a xenon lamp. The light emitted from the illumination light source 11 passes through a lens 101 having a desired beam diameter and directivity, is reflected by a mirror 102, and then passes through a spectral intensity and directivity adjusting illumination means 12. The spectral intensity and directivity adjusting illumination means 12 adjusts the illumination light to an optimum spectral illuminance distribution and makes light of each wavelength have a desired directivity. The light transmitted therethrough is reflected by the beam splitter 103, passes through the lens 104 and the mirror 105, passes through the reduction lens 5, and illuminates the alignment pattern 41 on the wafer 4. The light reflected by the alignment pattern 41 passes through the reduction lens 5 again, is reflected by the mirror 105, and enters the detection system 10x. The incoming detection light passes through the lens 104 and the beam splitter 103, and is then split into two by the wavelength selection beam splitter 106.
[0032]
This beam splitter has a wavelength of λ.₁To λ₂Reflects light up to λ₂To λ₃Transmit light up to. Since the detection light is separated into two wavelength bands and detected, the reduction lens 5 is designed and manufactured so that the highest resolution can be obtained with the exposure light.₁To λ₃This is because light having a wide wavelength range up to the above has large chromatic aberration, and it is difficult to manufacture a color correction lens that corrects this. That is, as shown in FIG.₁To λ₂Can be color corrected by the lenses 104 and 108,₂To λ₃Can be color corrected by 104 and 108 '. Λ separated by the wavelength selective beam splitter 106₁To λ₂The light up to this passes through the spectral transmittance and numerical aperture adjustment detecting means 13. The spectral transmittance and numerical aperture adjustment detecting means 13 makes it possible to detect with an optimum detection numerical aperture for each detection wavelength. The other wavelength separated by the wavelength selection beam splitter 106 is λ.₂To λ₃The reflected light is reflected by the mirror 107, and is similarly detected by the spectral transmittance and numerical aperture adjustment detecting means 13 'at the optimum detection numerical aperture at each wavelength. In this manner, the detection light of each detection wavelength band forms an image without color shift on the imaging surfaces of the one-dimensional or two-dimensional imaging devices 14 and 14 'by the color correction lenses 108 and 108'.
[0033]
The alignment patterns detected by the imaging devices 14 and 14 'are sent to the signal processing circuit 2, and the positions of the patterns are detected. If the position of the alignment pattern on the wafer is detected in this manner, this information and the position of the wafer table when each pattern is detected (not shown, usually by a laser length measuring device) The wafer stage is finely adjusted so as to perform correct overexposure based on the information of (measured), and the overexposure is executed. The above-mentioned spectral intensity and directivity adjusting illumination means 12 and spectral transmittance and numerical aperture adjustment detecting means 13 will be described later in detail with reference to embodiments, but the spectral intensity and directivity of illumination light adjusted by these means will be described. In addition, the optimum values of the spectral transmittance and the numerical aperture for detection generally differ depending on the structure and material properties of the wafer pattern to be detected. These optimum values may be calculated in advance off-line, and the values may be input from the input terminal 22 or, as shown in FIG. The value may be calculated.
[0034]
In the embodiment of FIG. 1, the spectral intensity and directivity adjusting illumination means 12 and the spectral transmittance and numerical aperture adjustment detecting means 13 are used at the same time, but the adjustment of the spectral intensity in 12 and the adjustment of the spectral transmittance in 13 are performed. One of the adjustments is sufficient.
[0035]
FIG. 4 shows an application of the pattern detection method of the present invention to an alignment evaluation device. 1 denote the same parts. On the wafer on the wafer table 7, for example, a resist pattern 412, which has been already exposed and developed, is formed as shown in FIG. That is, reference numeral 412 denotes a pattern formed by overlapping exposure. On the other hand, a pattern 411 is a pattern formed before the overlap exposure. In order to evaluate the superposition accuracy in the x direction, the deviation between the detection center Cp of the left and right patterns 411 and the detection center Cr of the left and right patterns 412 in FIG. good. However, as shown in FIG. 6, the normal pattern has a multilayer structure, and especially when the uppermost layer is an optically transparent layer, light having an optimum wavelength spectral characteristic is affected by multiple interference. Without illumination, the detected waveform will be erroneously detected as distortion. In this embodiment, since the pattern 411 and the pattern 412 have different layer structures, a wavelength which is totally optimal for both is selected.
[0036]
In FIG. 4, reference numeral 11 'denotes an illumination system, and the light emitted from the illumination system has a desired spectral intensity by the spectral intensity and directivity adjusting illumination means 12, and has directivity corresponding to each wavelength. That is, as will be described later, a secondary light source having a different size and intensity for each wavelength is created by the spectral intensity and directivity adjusting illumination means 12. After such light passes through the lens 101 'and is reflected by the beam splitter, an image of this secondary light source is formed on the pupil of the objective lens. The light passing through the pupil of the objective lens illuminates the pattern on the wafer telecentrically. The light reflected from the wafer passes through the objective lens 5 ', the beam splitter 105', and the image forming lens 109, and forms an image on the two-dimensional imaging device 14 '. Spectral transmittance and numerical aperture adjustment detecting means 13 are arranged between the imaging lens 109 and the two-dimensional imaging device 14 ', and detect with a numerical aperture according to the wavelength of the detection light.
[0037]
In the above embodiment, both the spectral intensity and directivity adjusting illumination means 12 and the spectral transmittance and numerical aperture adjustment detecting means 13 change the spectral intensity, but only one of them is sufficient. That is, for example, if the spectral intensity of the illumination light is adjusted optimally by the spectral intensity and directivity adjusting illumination means 12, the spectral transmittance and the numerical aperture adjustment detecting means 13 may have the same transmittance for each wavelength.
[0038]
If the two-dimensional pattern detected by the method shown in the present embodiment is detected by a two-dimensional imaging device, and the shift of the center of each of the patterns 411 and 412 in the x direction and the y direction is determined, the alignment is performed. Accuracy is required. Since the optimum illumination light spectral intensity, the numerical aperture of the optical system at the time of detection, or the directivity of the illumination light differ depending on the layer structure and material of the wafer, when these conditions are input from the input terminal 22, the processing circuit 2 ' The optimal illumination light spectral intensity and the numerical aperture of the optical system at the time of detection, or the directivity of the illumination light are calculated, and the spectral intensity and directivity adjustment illumination means 12 and the spectral transmittance and numerical aperture adjustment detection means 13 are driven. Thus, the illumination light spectral intensity and the numerical aperture of the optical system at the time of detection, or the directivity of the illumination light are realized.
[0039]
FIG. 7 shows a specific embodiment of the illumination means 12 for adjusting the spectral intensity and the directivity of the pattern detection apparatus of the present invention. FIG. 7 can also be an embodiment of the spectral transmittance and numerical aperture adjustment detecting means 13 of the present invention by partially changing the configuration as described later. Reference numeral 121 denotes a spectral unit such as a diffraction grating or a prism (not shown). The light emitted from the detection illumination light source 11 (11 ') becomes light 1010 with a certain degree of directivity increased in the y direction in FIG. The illumination light source for detection has a spectral intensity that includes the entire detection wavelength range, but does not have an optimal spectral intensity characteristic for detection. Therefore, the beams 1011A to 1011N split by the splitting means 121 are separated by the condensing lens 123 so as to condense the respective split lights on the spatial light transmittance modulator 125. When a diffraction grating is used, light with the shortest wavelength is condensed on 1012A, and light with the longest wavelength is condensed on 1012N. Since the zero-order transmitted light of the light applied to the spectral unit 121 is not used, the light is shielded by the light shielding plate 126. This light shielding plate may be arranged in front of the lens 123 as shown in FIG. 7, or may be arranged on the spatial light transmittance modulator 125.
[0040]
Therefore, as shown in FIG. 8A, the spatial light transmittance modulator 125 has a structure in which a light-transmitting portion 1251 is provided in a light-shielding portion 1252. To achieve this, the transmittance T (λ) is changed as shown in FIG. As a method of partially changing the light intensity in this way, for example, there is a liquid crystal shutter matrix, PZT, a phototomic material, or the like. As a simpler method, the object of the present invention can be achieved even if the change in the y direction is changed stepwise, so that the transmittance in the x direction changes gradually in the x direction as shown in FIG. The strip-shaped ND filter is configured to be movable in the X direction, and as shown in FIG. 19A, a number of strips arranged in the y direction are respectively controlled in the x direction, as shown in FIG. 8B. A desired transmittance may be obtained.
[0041]
Next, an embodiment of optimizing the directivity of illumination light for detecting each detection light at the same resolution as described above using Expression (1) will be described with reference to FIG. That is, in order to detect each detection light with the same resolution, the numerical aperture NA of the detection optical system is proportional to each detection wavelength λ, and further, the partial coherency of the illumination light at each wavelength with respect to this NA. Try to match. The spatial light transmittance modulation region 1253 of the spatial light transmittance modulator 125 'shown in FIG._BThen the short wavelength λ_A9 (A), the width of the transmission area of the spatial light transmittance modulator 125 'for illumination shown in FIG. 9 (A) so that the NA of the image detection is larger than that of the image detection and the partial coherence of the illumination is increased in proportion to this NA. Is wide. This width ratio is approximately equal to the wavelength ratio. By using a liquid crystal shutter matrix, PZT, or the like to drive the area of the spatial light transmittance modulation area 1253 with an electric signal, a desired NA (to be described later) for each wavelength and a partial coverage of illumination proportional to this are provided. To be detected by license.
[0042]
The light transmitted through the spatial light transmittance modulator 125 or 125 ′ is incident on a spectral unit 122 such as a diffraction grating or a prism (not shown) by the condenser lens 124 in FIG. 7. Since the surface of the light splitting unit 122 is conjugated to the surface of the light splitting unit 121 by the lenses 123 and 124, the light incident on the light splitting unit 121, that is, the light 1010 in which the directivity is increased to some extent in the y direction in FIG. Is reproduced on the spectral means 122. However, the intensity changes for each wavelength due to modulation by the spatial light transmittance modulator 125. As a result, the light that is transmitted and diffracted through the spectral unit 122 illuminates the wafer 4 with a desired partial coherence of illumination and a spectral illuminance distribution.
[0043]
FIG. 7 illustrates the case where 12 is used for the illumination system. However, by using 13 for the detection system with the same configuration as 12 in FIG. 7, a desired detection NA is given to each of the above wavelengths, and detection at each wavelength is performed. It allows the resolution of the images to be identical. That is, in this case, the reference numeral 13 in FIG. 7 is arranged at, for example, the positions 13 and 13 'in the embodiment of FIG. 1, and the light reflected by the pattern 41 on the wafer reflects the reduction lens 5, the mirror 105, and the beam. After passing through the splitter 103, reflected by the beam splitter 106 or the mirror 107, imaged on the spectroscopic means 121 in FIG. 7, and diffracted at different diffraction angles for each wavelength, the lens 123, the spatial light transmittance modulation The image of the pattern 41 is formed again on the spectroscopic means 122 by the lens 124 through the container 125.
[0044]
In the spatial light transmittance modulator 125, as shown in FIG. 9A, the width of the light transmission region 1251 changes so as to satisfy Expression (1) for each wavelength, that is, along the height direction y. However, when the transmittance is changed for each wavelength in the illumination system as shown in FIG. 8 (B) and FIG. 9 (B), the spatial light transmittance modulator 125 ′ in the detection system is changed for each wavelength. Is not necessarily required. The pattern re-imaged on the spectroscopic means 122 by the lens 124 enters the spectroscopic means 122 at different incident angles for each wavelength and forms an image, but the diffracted light 1020 transmitted therethrough has the same light of each wavelength. Follow the light path. The diffracted transmitted light re-images the image on the spectroscopic unit 122 on the imaging surfaces of the one-dimensional or two-dimensional imaging devices 14 and 14 'by the color correction lens 108 or 108' of FIG.
[0045]
The color correction lens 108 or 108 'corrects various aberrations including chromatic aberration with respect to the light within the respective wavelength bandwidths, and the images of the light of the respective bandwidths are one-dimensional or two-dimensional imaging devices 14 and 14'. Image without distortion and color shift.
[0046]
Next, a method for specifically detecting a pattern using the configuration of the embodiment of the pattern detecting apparatus described above will be described. Since the pattern detection target has a layer structure as shown in FIG. 3, when the uppermost layer of the pattern is a photoresist, the thickness of the photoresist is as shown in FIG.
[0047]
Since the photoresist is usually applied by a spin coater, if the center of the wafer is at the upper left of FIG. 3, the flow of the photoresist flows from left to right. As a result, as shown in FIG. 10A, the thickness of the photoresist is larger at the right end than at the left end of the pattern edge. As a result, the range of the thickness of the photoresist becomes t as shown in FIG._B1To t_B2Up to t_T1To t_T2There will be a range of thickness up to.
[0048]
If the resist structure, the layer structure, and the optical constants (such as the complex refractive index) of the material of the resist and the underlying pattern are known, the interference calculation can be performed using these conditions as parameters as shown in FIG. Do. That is, when the wavelength is changed with respect to the above parameters, the calculation is made as to what the interference intensity will be. From the obtained results, it is necessary to irradiate with any spectral illumination intensity distribution, and the resist thickness range of 10 (A), that is, t_B1To t_B2And t_T1To t_T2The spectral illumination intensity is determined by calculating whether the change in the interference intensity is small and approaches a constant value. Spectral illumination intensity distribution I determined in this way_i(Λ) is, for example, as shown in FIG. This distribution is merely a spectral illuminance distribution of the light illuminating the wafer, and in fact, the spectral intensity distribution of the light source and the spectral transmittance distribution of the optical system are not constant, and thus need to be corrected.
[0049]
That is, for example, the spectral intensity distribution I of the light source_iil(Λ) is as shown in FIG. 11B, and the spectral transmittance distribution I of the optical system is_OSIf (λ) is as shown in FIG. 11C, the transmittance I (λ) for each wavelength of the spatial light transmittance modulator 125 is a value obtained by the following equation as shown in FIG. 11D. Need to be
[0050]
(Equation 2)

[0051]
If the spatial light transmittance modulator 125 is driven so as to be I (λ) determined in this way, the range of the resist thickness as shown in FIG._B1To t_B2And t_T1To t_T2The change in the interference intensity is small until then, and the waveform of the detection pattern is not affected by the unevenness in the application of the resist, is symmetrical, and a highly accurate pattern position detection is realized.
[0052]
An example in which the pattern detecting method of the present invention is applied to a pattern position detecting device will be described with reference to FIG. The pattern position detecting apparatus according to the present embodiment is applied to an off-axis alignment detecting system of a stepper or a detecting system for alignment evaluation. The detection mark 41 on the wafer 4 has a structure as shown in FIG. Illumination light 1100 is illuminated on the detection objective lens 5 'through the beam splitter 51, the imaging lens 51 and the objective lens 5'. The reflected light returns to the light path of the illumination in the reverse direction and is reflected by the beam splitter to become detection light 1200. The detection light forms an image of the detection pattern on the spectral means 121. Light that passes straight through the light splitting means is shielded by the light shielding plate 126.
[0053]
On the other hand, the light diffracted by the spectroscopic means changes its diffraction angle for each wavelength, and the light of a shorter wavelength_{λ 1}And long wavelength light is D_{λ n}Focus on The light of each wavelength transmitted through the condenser lens 123 becomes parallel to each other and passes through a numerical aperture determining aperture 1151. As a result, the NA changes according to the wavelength as described above. The light that has passed through the numerical aperture determining aperture 1151 enters the cylindrical lens 124 '. The cylindrical lens 124 ', which is a non-rotationally symmetric image forming system, has an image forming function in the pattern position detecting direction x, and has no image forming function in a direction perpendicular thereto. Accordingly, on the imaging surface 140 of the imaging device 14 ", a spectral one-dimensional image in a state of being spectrally separated in the y direction, that is, an image is formed in the x direction in an imaged state in which the pattern on the wafer is enlarged. This image is formed side by side in the vertical direction y for each color.
[0054]
That is, wavelength λ₁To λ_nThe image of light of each wavelength up to 140 P on the imaging surface_{λ 1}To P_{λ n}Is formed at the position. Therefore, P is displayed on the TV monitor-210 of the signal processing device 21 in accordance with each wavelength._{λ 1}’……, P_{λ i}’……, P_{λ n}The image of each wavelength is formed at the position of ', and a detection signal at each wavelength is obtained.
[0055]
A method of detecting a pattern in the signal processing circuit 21 using the signals separated and detected for each wavelength as described above will be described below. Λ of the detected waveform of each wavelength detected by the imaging device 14 ″₁, Λ_i, And λ_nFor example, the detection waveform at each wavelength of I is shown in FIG._{λ 1}(X), I_{λ i}(X), I_{λ n}(X). The levels of these detected waveforms are based on the spectral intensity distribution I of the light source._ill(Λ_i) And the spectral transmittance distribution I of the optical system_OS(Λ_i). Further, as described above, the optimum spectral illumination intensity I is determined from the resist film thickness of each part of the pattern, the underlying structure, and the material._i(Λ_i), The above-mentioned detection waveform I of each wavelength is used._{λ 1}(X), I_{λ i}(X), I_{λ n}(X) shows a correction value α (λ_i) May be used as the optimum detection waveform.
[0056]
(Equation 3)

[0057]
Therefore, the optimal detection waveform I (x) finally obtained is obtained by the following equation.
[0058]
(Equation 4)

[0059]
The detection waveform thus obtained is as shown in FIG.
[0060]
Next, another embodiment relating to the signal processing of the present invention will be described. The detection optical system in FIG. 16 uses an optical system in which chromatic aberration has been corrected for the detection light. However, since a normal microscope objective lens is used for visual observation, color bleeding occurs when an object is visually observed. If not, it will withstand the specifications. However, in the evaluation of alignment, it is necessary to evaluate the alignment accuracy with an accuracy of about 10 nm. For this reason, even if the chromatic aberration is invisible to the naked eye, it affects the evaluation accuracy. Therefore, it is important to evaluate how much chromatic aberration remains in the state where the optical system of FIG. 16 is assembled, correct this residual chromatic aberration at the time of alignment accuracy evaluation, and perform detection and evaluation when the chromatic aberration is 0. become.
[0061]
This is realized in the present embodiment. That is, in a state where the optical system of FIG. 16 is assembled, an interference filter or the like is inserted in front of the illumination light source or the imaging surface of the imaging device, and the chromatic aberration when a single wavelength is detected is evaluated. For example, a concave pattern of a predetermined size is formed on a Si wafer, and this is detected with various single wavelengths or light with a narrow bandwidth, and the detected position and the pattern width or pattern interval are measured. Keep it. FIG. 18A is an example of the result of evaluating the chromatic aberration in this manner. That is, the horizontal axis represents wavelength, and the vertical axis represents lateral magnification. Thus, the wavelength λ₁, Λ_i, And λ_nIf the lateral magnification is different at the same wavelength, the detection signal at each wavelength detected in the embodiment of FIG. 16 has a pattern width W as shown in FIG._{λ 1}, W_{λ i}, And W_{λ n}Will be different.
[0062]
However, since the waveforms at these wavelengths are obtained independently, the lateral color magnification M (λ₁), M (λ_i), M (λ_n) Can be used to correct the magnification. FIG. 18C shows the result of this correction. That is, the corrected result has the same pattern width or interval at any wavelength, and the detected positions can be made to coincide.
[0063]
If the correction of the spectral illumination intensity and the like described with reference to FIG. 17 is performed after the magnification is corrected in this manner, an ideal pattern detection is realized as shown in FIG. That is, in this embodiment, the chromatic aberration that cannot be completely removed by the optical system is spectrally detected to separate and detect the waveform at each wavelength, and the waveform of each wavelength is corrected (corresponding to the color magnification) at the detection position coordinates. Correction, offset of the detection center for each color (positional deviation), correction of positional deviation for each color due to image distortion for each color, and the like, and it is possible to perform level correction for each detection signal. .
[0064]
In addition to the above-described correction, the pattern detection device of FIG. 16 can obtain a detection waveform for each wavelength even when illumination unevenness occurs for each wavelength of the illumination light. This uneven illumination can be easily corrected.
[0065]
Further, the advantages of the pattern detecting device of FIG. 16 will be described. As shown in FIG. 16 (B) and FIG. 10, the thickness of the resist is different between the periphery of the concave pattern and the pattern portion (the concave portion) as shown in FIG. 10 (A). There is an optimum illumination spectral intensity for the peripheral resist thickness, which does not always match that of the pattern portion. In such a case, since the detection waveform is obtained for each wavelength in the present embodiment, α (λ_i) Is used. That is, α (λ_i) As a function of the coordinate x, α (λ_i) Can be set to an optimum value, a detection waveform without distortion can be obtained, and detection can be performed with the highest accuracy. Note that α (λ_i) To a function α (λ_i, X), for example, as shown in FIG._L, X_RIs discontinuous at If such a discontinuous portion occurs, a discontinuous portion or distortion occurs in the detected waveform. Therefore, a process for smoothly continuousizing the discontinuous portion is performed.
[0066]
In the embodiment of FIG. 16, only the optical system for detecting the pattern in the x direction is displayed. However, in the case of detecting the xy two axes, a beam splitter is inserted between the spectral means 121 and the imaging lens 51. , X-direction (a configuration for detecting the y-direction) may be provided.
[0067]
FIG. 7 illustrates the spectral intensity and directivity adjusting illumination means 12 and the spectral transmittance and numerical aperture adjustment detecting means 13 of the pattern detecting apparatus of the present invention, and FIG. 16 shows the spectral transmittance and numerical aperture adjustment detecting means. Although the means 13 has been described, there is an optimum spectral directivity for each wavelength set by the spectral transmittance and the numerical aperture adjustment detecting means. That is, as described above, in order to provide the same resolution for any detection wavelength, the NA is set so that the resolution L shown in the equation (1) becomes almost the same value for all detected wavelengths. Partial coherence σ of illumination determined for this NA (λ) (an aperture that determines the numerical aperture for detection when the object is flat, ie, the object to be detected on 1251 (The ratio of the width of the reflected light to the width of the opening) is substantially constant regardless of the wavelength. By doing so, the MTF (Modulation Transfer Function) is the same for any wavelength, and a wavelength composite waveform with little distortion can be obtained.
[0068]
FIG. 20 shows an embodiment of the present invention in which a two-dimensional pattern is not separated by xy and two directions are independently detected, but not two directions.
[0069]
FIG. 20 shows an embodiment of an illumination system used for two-dimensional simultaneous detection. The detection light source 11 ′ causes the light emitted from the mercury lamp or xenon lamp 111 to pass through the mirror- 113 by the ellipsoidal mirror 112 and to be condensed on the incident end of the optical fiber 121. The optical fiber 121 is branched into a plurality of fiber bundles on the way. The light emitting ends 1211, 1212, 1213, 1214, and 1215 of the plurality of branch fibers are shifted from each other in the optical axis direction as shown in the drawing. A wavelength selection filter and directivity adjusting illumination means 1221, 1222, 1223, 1224, and 1225 are provided immediately after the light emitting end of the fiber.
[0070]
That is, 1221 transmits light having a wavelength of 640 nm to 700 nm and has a desired light intensity by using an ND filter or the like in an overlapping manner. As shown in FIG. 20B, an opening 1221A indicated by a solid line is opened inside the light shielding portion 1220 of the wavelength selection filter, the directivity adjusting illumination means 1221, and the like.
[0071]
Similarly, 1212 transmits light of 580 nm to 640 nm. In the following, the desired light intensity is obtained by using a liquid crystal shutter or the like 1213 for transmitting light of 520 nm to 580 nm, 1214 for light of 460 nm to 520 nm, 1215 for light of 400 nm to 460 nm and changing the transmittance with an ND filter or an electric signal. To For example, in 1215, an opening 1225A indicated by a dotted line in FIG. The light that has passed through each wavelength selection filter and the directivity adjusting illumination means enters a wavelength selection beam splitter 123.
[0072]
The reflection surface 1231 is a total reflection. For example, the 1233 surface reflects light having a wavelength of 580 nm or less and transmits light having a wavelength of 580 nm or more. The light that has passed through such a wavelength selection beam splitter 123 is used as illumination light for the detection optical system. For example, taking the case where this illumination system 12 'is used in the embodiment of FIG. 4, the apertures 1221A to 1225A are imaged by the lens 101' on the entrance pupil of the pattern detection lens 5 ', and The pattern to be detected can be illuminated with color illuminating light having proportional illumination directivity.
[0073]
In this manner, it is possible to separate the light into each wavelength band, to set the light in each wavelength band to a desired intensity ratio, and to irradiate the test object with the light in each wavelength band to a desired illumination directivity. . In this embodiment, since the directivity of the illumination has a rotational symmetry about the optical axis, the desired illumination is obtained regardless of whether the pattern of the test object is one-dimensional or two-dimensional. Can be realized.
[0074]
Next, FIG. 21 shows an embodiment of a detection optical system used for two-dimensional simultaneous detection.
[0075]
The reflected light from the pattern of the wafer illuminated using the spectral intensity and directivity adjusting means 12 'of the embodiment of FIG. 20 passes through the detection objective lens shown in FIG. You. Before the image is detected by the two-dimensional imaging device 14 ', it passes through the spectral transmittance and numerical aperture adjustment detecting means 13 "shown in FIG. 21. This spectral transmittance and numerical aperture adjustment detecting means 13" is implemented as shown in FIG. When applied to the example, the spectral transmittance and numerical aperture adjustment detecting means 1331, 1332, 1333, 1334, and 1335 of each of the above-mentioned wavelength bands are arranged almost at the rear focal point of the imaging lens 109. Up to this point, each wavelength is separated by the beam splitter 131. This beam splitter is basically the same as 123 in FIG.
[0076]
The spectral transmittance and numerical aperture adjustment detecting means 1331, 1332, 1333, 1334, and 1335 of each wavelength band have an opening 1331A (３１1335A) at the center of the light shielding portion 1330 as shown in FIG. It is open, and the numerical aperture NA for detection is determined by this aperture diameter. The lights transmitted therethrough are again synthesized by the beam splitter 132 and form an image on the two-dimensional imaging device 14 '. Note that the beam splitter 132 is different from the beam splitters 131 and 123. For example, the 1323 surface transmits light having a wavelength of 520 nm or less and reflects light having a wavelength of 520 nm or more. The diameter of the aperture 1331A (〜1335A) is determined so that the resolution in each wavelength band is substantially equal, for example, so that the resolution L shown in Expression 1 is substantially equal in the center of each wavelength band. This makes it possible to detect a pattern with substantially the same resolution at each wavelength and with optimum spectral illumination characteristics according to the structure and material of the object.
[0077]
FIG. 22 (A) shows an embodiment of the present invention, which is different from the embodiment shown in FIG. 21 and is separated into each wavelength band, and the detection light reflected from the pattern to be detected is separated into each wavelength band separately. This is detected by a two-dimensional imaging device. The light separated into each wavelength band by the beam splitter 131 is subjected to a desired detection numerical aperture NA for each wavelength band by numerical aperture adjustment detecting means 1331 ', 1332', 1333 ', 1334' and 1335 '. The images are formed on the two-dimensional imaging devices 141 ′, 142 ′, 143 ′, 144 ′, and 145 ′ by the imaging lenses 1091, 1092, 1093, 1094, and 1095. The detection signal of each wavelength band obtained by each two-dimensional imaging device is input to the signal processing circuit 2 '.
[0078]
The coordinates (x ′, y ′) of an arbitrary point P on the imaging surface of each imaging device shown in FIG. 22B are affected by aberrations such as chromatic aberration of the detection optical system, and the detection optical system is not used. Image point P in case of ideal optical system with aberration₀(X, y). Let the ith wavelength band be λ_iAnd the wavelength band λ_iPoint P (x ', y') and point P in the imaging system₀It is assumed that the relationship of (x, y) can be expressed by Expression 5.
[0079]
(Equation 5)

[0080]
Here, (x, y) (x ′, y ′) is a position vector on two-dimensional coordinates, and p_{λ i}Is the wavelength band λ_iThis is a function of coordinate transformation that differs for each, that is, a function representing distortion due to aberration. Function p_{λ i}(X ', y') can be obtained, for example, by forming a regular grid pattern on the wafer, detecting using this pattern, and detecting the position signal of the detected image.
[0081]
FIG. 23 shows another embodiment for detecting a two-dimensional pattern. Using the illumination light source shown in FIG. 20 (not shown), the emission end of the fiber 121 'emits illumination light without branching as in the embodiment of FIG. The emitted light passes through the directivity adjusting unit 122 'shown in FIG. 23B, and passes through the wavelength selecting unit 1255. The directivity adjusting means 122 'sets the above-described illumination directivity to a desired value for each wavelength. For this reason, as shown in FIG. 23B, a desired region 1221A '(wavelength is λ_iIs set to a desired transmittance, and the outside is set to a transmittance of 0.
[0082]
As shown in FIG. 23 (D), band-pass filters 12551 to 12558 having slightly different transmission wavelength bands are arranged on the disk in the wavelength selecting means 1255 in the circumferential direction. For example, a wavelength band λ obtained by equally dividing 400 nm to 700 nm into eight.₁~ Λ₈Is obtained by rotating the disk. The illumination light obtained in this manner is applied to a test object such as a wafer in the same manner as in the above embodiment, and the obtained reflected light is used as a numerical aperture adjustment detecting means 133 'having the same function as the above 122'. Lead to. 133 'is at a position conjugate with the pupil of the detection optical system, and a desired detection numerical aperture NA can be obtained by setting the inside of 1331A' in the light transmission area 1330 'to the transmission state and the outside to the light shielding state.
[0083]
Wavelength band λ₁Is completed and the detection information I_{λ i}If (x ′, y ′) is stored in the signal processing circuit 2 ″, the disk of the wavelength selection unit 1255 is rotated, and the wavelength band λ₂Is driven, and the directivity adjusting means 122 'and the numerical aperture adjustment detecting means 133' are driven to set, detect, and store the determined conditions in this wavelength band. The above operation is sequentially performed in the wavelength band λ₈The storage of the detection signals in all the wavelength bands is completed. A specific example of how to use the obtained signal will be described below.
[0084]
FIG. 24 shows a method for obtaining the deviations .DELTA.x and .DELTA.y in the x and y directions of 411 and 412 of the two-dimensional pattern 41 'shown in FIG. 5 obtained by the method described with reference to FIGS. This will be described with reference to FIG. FIG. 24 shows the flow of processing of the detection signal and the contents of the processing. Each block can be individually realized by a dedicated electronic circuit, but the portion where the processing time is short is processed by a common microcomputer. Wavelength band λ_iThe detection signal obtained by the detection of_{λ i}(X ', y'). As described above, the patterns 411 and 412 have different structures and materials, so that the optimum spectral illuminance distribution for detection is different.
[0085]
Embodiments In FIGS. 20, 22, and 23, since a detection waveform can be obtained independently in each wavelength band, the obtained signals are added to each of the patterns 411 and 412 and the detection wavelength so that the optimum spectral illuminance distribution is obtained. Band λ_iConstant α for each_{1 λ i}, Α_{2 λ i}And the above-mentioned aberration correction is performed by 221 and 221 '. The obtained corrected waveforms are shown by 221 and 221 'in FIG._{1 λ i}’(X, y) and I_{2 λ i}′ (X, y). The corrected waveforms obtained at the respective wavelengths are added to the patterns 411 and 412 over all the wavelengths at 222 and 222 ', and the intensity waveforms are added for each position coordinate to obtain the ideal I.₁(X, y) and I₂(X, y) is obtained.
[0086]
Using the obtained waveform, for example, the center position C of each of the patterns 411 and 412 in the x and y directions using an algorithm of symmetric pattern matching._1x, C_1y, C_2xAnd C_2yIs obtained by 223 and 223 '. In FIG. 24, f_xAnd f_yRepresents an operation function for obtaining a pattern center (for example, an operation function for symmetric pattern matching). From the centers of the patterns 411 and 412 thus obtained in the x and y directions, the displacements Δx and Δy in the x and y directions of both patterns are obtained by 224. Next, an example of how to use the thus obtained positional deviation information will be described.
[0087]
FIG. 25 is an embodiment diagram showing the exposure system of the present invention.1Denotes an exposure apparatus, for example, a reduction exposure apparatus shown in FIG. Reference numeral 2 denotes a control circuit for controlling the exposure apparatus. Reference numeral 10 denotes, for example, the pattern detection device or the alignment evaluation device shown in FIG. 2 'is a control circuit of this device. Reference numerals 91, 92, and 93 denote a resist coating device, a developing device, and an etching device, respectively. A control circuit 8 controls these photolithography systems. In the figure, the flow indicated by the dotted line is the flow of the wafer, and the flow indicated by the solid line is the flow of the signal.
[0088]
The wafer coated with the resist by the resist coating device 91 is an exposure device.1It is carried in. Since an alignment pattern is recorded on the loaded wafer for each exposure chip as shown in FIG. 2, this alignment pattern 41 (41 ') is detected by the detection system 10X (10Y) through the reduction lens 5. I do. The wafer stage 7 is driven based on the position information thus detected, the position information of the laser stage length measuring device (not shown in FIG. 1) and the position information of the reticle 3, and the alignment is performed. After that, exposure is performed.
[0089]
Such exposure is performed in a step-and-repeat manner over the entire surface of the wafer, and the circuit pattern drawn on the reticle 3 is exposed over the entire surface of the wafer. As described above with reference to FIG. 1 and the embodiment of the detection system 10X, in the exposure apparatus equipped with the pattern detection method of the present invention, correct alignment pattern detection is performed, and accurate overlay exposure is performed. Has been realized. The exposed wafer is sent to the developing device 92 and developed. For example, a pattern 411 shown in FIG. 5 is recorded on the developed wafer before exposure in order to evaluate overlay exposure. Since a new resist pattern 412 is exposed on this pattern 411, the relative positions of these patterns 411 and 412 are determined by using ten alignment evaluation devices as described in the previous embodiment. By detecting, the accuracy of the overlay exposure is evaluated. The evaluation result is output from the control circuit 2 ', and the output 23 is sent to the control circuit 8.
[0090]
The control circuit 8 includes a resist coating machine 91 and an exposure device1If the evaluation result of the overexposure is not within the target accuracy, the cause is analyzed. If the cause lies in the resist coating machine, control is made to optimize the conditions for resist coating, for example, the conditions such as the spin coat rotation speed, rotation acceleration / deceleration conditions, and the resist solvent vapor pressure or temperature of the coating machine. If the cause of the poor overlay evaluation result is in the exposure apparatus, it is determined where in the exposure apparatus the cause is.
[0091]
Originally, an abnormality check function, an alignment detection result, position information at the time of exposure, and the like are recorded in the exposure apparatus, and the state of the apparatus is constantly monitored and recorded. The information is sent to the control circuit 8, and using this information and the information sent from the alignment evaluation device 10, an exposure apparatus is used.1It is determined whether there is a cause in the alignment detection system. This determination can be largely performed by the control circuit 8 having a learning function. If the above-described illumination means and detection means are provided in the alignment detection system of the exposure apparatus as in the present embodiment, superimposition exposure with considerable accuracy should be realized, but higher accuracy can be achieved. In the case where overlay exposure is required or the exposure apparatus does not have the above-mentioned illumination means and detection means, feedback is provided to the exposure apparatus based on the alignment evaluation result, and high alignment is performed in the subsequent exposure. It is necessary to obtain accuracy. Two methods are effective as this feedback method.
[0092]
In the first method, x, y, and θ obtained as a result of the alignment evaluation (θ is the position x of two x-direction detection patterns at two positions r away from the exposure chip.₁, X₂Θ = (x₁-X₂The values Δx, Δy, Δθ of the overlay error in the directions) / r) are given as offset values at the next exposure, and the correction is applied by software. That is, the offset value when this wafer is exposed is, for example, Δx₀, Δy₀, Δθ₀, The offset value Δx at the time of the next exposure with this feedback multiplied.₁, Δy₁, Δθ₁To Δx₁= Δx₀-Δx, Δy₁= Δy₀−Δy, Δθ₁= Δθ₀Exposure is performed at −Δθ. By doing so, the overlay accuracy after feedback is greatly improved.
[0093]
In the second method, conditions such as the spectral illuminance distribution and illumination directivity of the above-described illumination system of the alignment detection system of the exposure apparatus, and the numerical aperture for detection of each wavelength of the detection optical system are changed so as to be more optimal. . By feeding back the alignment evaluation result to an exposure apparatus or a resist coating apparatus using any of the above methods, the overlay exposure accuracy is greatly improved.
[0094]
Next, an embodiment of the exposure system of the present invention will be described with reference to FIG. 25 and FIG. 25 represent the same items. The dotted arrows in the figure indicate the flow of the wafer, but since almost all the cases are shown together, each flow is numbered starting with F for easy understanding of the following description. ing. Solid arrows represent the flow of signal information. Hereinafter, a plurality of embodiments (A) to (E) will be described with reference to FIG.
[0095]
(Example A)
Wafer flow: F11 → F12 → F21 → F22 → F3 → F4 → F5
The wafer alignment pattern (resist not coated) formed in the previous exposure and development etching steps is evaluated by the alignment evaluation device 10 to evaluate the symmetry of the wafer pattern. Usually, the symmetry is good. In this case, the evaluation is performed again by the alignment evaluation device 10 with the subsequent resist applied. If the result of the detection of the alignment pattern is asymmetric, it can be determined that the application of the resist is asymmetric.
[0096]
That is, in this case, as described above, the above-described exposure apparatus is used so that the detected waveform is symmetric with respect to the asymmetry of the resist coating.1Exposure apparatus using spectral illuminance and directivity adjustment illumination means 12 and spectral transmittance and numerical aperture adjustment detection means 13 of the alignment detection system to detect under optimum conditions1Feed forward to In this way, the exposure apparatus1In the alignment detection performed prior to the exposure, accurate overlay exposure is realized without being affected by the unevenness of the resist coating.
[0097]
In this case, it is not necessary to evaluate the alignment result again using the alignment evaluation device after the exposure and development. However, for the purpose of confirmation, evaluation F41 → F42 of the alignment result may be performed instead of F4. If the symmetry of the wafer pattern is poor as a result of evaluation by the alignment evaluation device 10 after the above-mentioned F11, the result is sent to the control device 8 and fed back to the etching device (not shown), or the asymmetry is obtained. Exposure equipment1Feed forward to
[0098]
(Example A ')
Wafer flow: F11 → F12 → F21 → F22 → F3 → F4 → F5
The flow of the wafer is the same as that of the embodiment A.
[0099]
In this embodiment, the exposure apparatus1It is assumed that the above-mentioned means of the present invention is not applied to the alignment detection system, and that the monochromatic light obtained by a mercury lamp or laser light is detected as an illumination light source. Even in such a case, by using this embodiment, it is possible to perform exposure with high alignment accuracy.
[0100]
The following shows how to achieve this.
[0101]
First, the conditions of the detection system of the alignment evaluation apparatus 10 are set to be the same as those of the alignment detection system of the exposure apparatus. This is because the alignment evaluation device 10 can control the spectral illuminance, the directivity of illumination, and the numerical aperture for detection, so that the same conditions as those of the alignment detection system of the exposure device can be easily realized. . In such a state (the same conditions as the alignment detection system of the exposure apparatus), the position of the alignment pattern of the wafer is detected.
[0102]
The result is, for example, (x₁, Y₁Next, while keeping the position of the wafer fixed, the optical system of the alignment evaluation device is changed so as to satisfy the above-described optimum conditions for detection, and the position of the alignment pattern of the wafer is detected. The result is (x₀, Y₀).
[0103]
Using these detection results, the offset value Δx = x₀-X₁, Δy = y₀-Y₁, And adding this value to the alignment detection result at the time of the next exposure, it is possible to perform the overlay exposure with high accuracy. The flow after exposure is the same as in Example A.
[0104]
(Example B)
Wafer flow: F1 → F21 → F12 → F21 → F22 → F3 → F4 → F5
When the symmetry of the pattern formed on the wafer is kept stable, it is not necessary to evaluate the alignment pattern before applying the resist. In this case, the evaluation after the application of the resist is performed by the alignment evaluation device, and the illumination and detection conditions of the alignment detection system of the exposure device are optimized as described above. The subsequent flow is the same as that of the embodiment A.
[0105]
(Example B ')
Wafer flow: F1 → F21 → F12 → F21 → F22 → F3 → F4 → F5
The flow of the wafer is the same as in the embodiment B.
[0106]
In this embodiment, as in the embodiment A ', the exposure apparatus1The above-mentioned means of the present invention is not applied to the alignment detection system described above, and for example, a monochromatic light obtained by a mercury lamp or laser light is detected as an illumination light source. Similarly to Embodiment A ', the alignment evaluation apparatus 10 determines an offset for alignment detection of the exposure apparatus from the result of detection under the same detection conditions as the alignment detection system of the exposure apparatus and the result of detection under the optimum conditions, and performs overlay exposure.
[0107]
(Example C)
Wafer flow: F1 → F2 → F3 → F41 → F32 → F4 → F5
Since this embodiment is the same as FIG. 25, the description will be omitted.
[0108]
(Example D)
Wafer flow: F1 → F2 → F3 → F41 → F42 → F51 → F52
After the development, the alignment evaluation device 10 evaluates the alignment accuracy of the pattern 411 formed before the exposure and the resist pattern 412 formed by the exposure, as shown in FIG. After the evaluation, etching is performed, and the accuracy of alignment between the etching pattern formed by the resist mask 412 and the pattern 411 is evaluated. If these two evaluation results match, there is no problem, but if they do not match, this mismatch amount is added as an alignment offset value when exposing the next wafer, so that the alignment offset generated due to etching can be reduced. Correction can be performed, and overlay exposure of the pattern of the next wafer can be realized with high accuracy.
[0109]
(Example E)
Wafer flow: F1 → F2 → F31 → F32 → F41 → F42 → F5
The wafer after exposure and before development is evaluated by an alignment evaluation device. This evaluation evaluates the accuracy of alignment between the latent image formed on the resist pattern after exposure and the pattern of the base already formed. Since the shift amount obtained in this evaluation is obtained before developing this wafer, this result can be used immediately as an alignment offset value at the time of exposure of the next wafer, so that the throughput does not drop, Overlay exposure is realized with high accuracy.
[0110]
【The invention's effect】
According to the present invention, alignment in exposure of a semiconductor circuit can be performed with high precision without depending on coating unevenness of a resist applied on a wafer or a film structure of a base pattern, and a circuit having a high mounting density can be obtained at a high speed. It has become possible to produce with dome. In addition, the measurement of the dimensions and intervals of the formed patterns, which are performed in the production of semiconductor circuits, can be performed with high accuracy, and the production yield can be improved.
[Brief description of the drawings]
FIG. 1 is a block diagram showing a pattern detecting method and apparatus according to the present invention and a projection exposure apparatus using the same.
FIG. 2 is a plan view of a circuit and a detection pattern on a wafer as a test object.
FIG. 3 is a perspective view showing a structure of a detection pattern.
FIG. 4 is a block diagram of a pattern detection method and apparatus according to the present invention.
FIG. 5 is a plan view of a pattern of a test object.
FIG. 6 is a sectional view of a pattern of a test object.
FIG. 7 is a cross-sectional view showing a spectral intensity and directivity adjusting illumination unit and a spectral transmittance and numerical aperture adjustment detecting unit according to the present invention.
FIG. 8 is a plan view of a first spatial light transmittance modulator of the present invention and a graph showing the transmittance thereof.
FIG. 9 is a plan view of a second spatial light transmittance modulator of the present invention and a graph showing the transmittance thereof.
FIG. 10A shows the thickness of a resist applied by a spin coater.
(B) Multiple interference intensity against resist thickness
(C) Example of desired spectral illumination intensity
FIG. 11 (A) Example of desired spectral illumination intensity
(B) Spectral intensity of light source
(C) Spectral transmittance of optical system
(D) Transmittance of spatial light transmittance modulator of spectral intensity and directivity adjusting illumination means (or spectral transmittance and numerical aperture adjustment detecting means)
FIG. 12 is a block diagram showing a process of determining a spectral transmittance according to the present invention.
FIG. 13 is a diagram showing a cross-sectional structure in the case where resist coating unevenness is present and a detection waveform when detection is performed using conventional single-wavelength light.
FIG. 14 is a diagram showing a detection waveform detected by conventional white light detection when the spectral reflectance of a base material has a strong wavelength dependency.
FIG. 15 is a view showing the operation of the present invention.
FIG. 16 is a block diagram showing a pattern detection method and apparatus according to the present invention.
FIG. 17 is a view showing a detection waveform.
FIG. 18 is a diagram illustrating chromatic aberration correction according to the present invention.
FIG. 19 is a plan view showing a spatial light transmittance modulator constituted by ND filters on strips according to the present invention, and a graph showing the transmittance thereof.
FIG. 20 is a block diagram showing an illumination system of the present invention when detecting a two-dimensional pattern.
FIG. 21 is a block diagram showing a detection system of the present invention when detecting a two-dimensional pattern.
FIG. 22 is a block diagram showing a detection system of the present invention when detecting a two-dimensional pattern.
FIG. 23 is a block diagram showing a detection system of a two-dimensional pattern.
FIG. 24 is a diagram showing a processing flow of the present invention when a two-dimensional pattern is detected.
FIG. 25 is a block diagram showing an exposure system of the present invention.
FIG. 26 is a block diagram showing an exposure system of the present invention.
[Explanation of symbols]
DESCRIPTION OF SYMBOLS 1 ... Exposure apparatus, 2 ... Signal processing circuit, 4 ... Wafer, 5 ... Reduction projection lens, 5 '... Objective lens, 8 ... Control device, 10 ... Pattern detection device, 11 ... Light source, 14 ... Pattern detection Means 22, signal input terminal, 41 detection pattern, 91 resist coating device, 92 developing device, 93 etching device, 121 spectral device, 124 'cylindrical lens, 1151 spatial light transmittance modulator .

Claims (6)

A resist is coated on the surface of the substrate on which the alignment pattern is formed, and the alignment pattern of the substrate coated with the resist is emitted from a light source having a wide wavelength width or a light source that effectively includes a plurality of wavelengths, and the spectral intensity of the light source is obtained. Irradiate light with different intensities so as to have a desired intensity distribution according to the wavelength including the influence of the distribution, image the alignment pattern irradiated with the light , and transfer it to the substrate and the substrate. Perform alignment with the reticle on which the pattern is formed, transfer the pattern of the reticle to the aligned substrate, expose the pattern, develop the exposed substrate by transferring the pattern, develop the developed pattern The relative position with respect to the alignment pattern was detected, and the accuracy of the overexposure of the developed pattern and the alignment pattern was determined. If the accuracy of the exposure is lower than a preset reference value, the cause is analyzed, and either an apparatus for applying a resist based on the analysis result or an exposure apparatus for transferring and exposing the pattern of the reticle. The overlapping exposure method characterized by adjusting the following. 2. The overlay exposure method according to claim 1, wherein an offset amount of alignment between the substrate and the reticle in the exposure apparatus is adjusted based on a result of the analysis. A resist is coated on the surface of the substrate on which the alignment pattern is formed, and the alignment pattern of the substrate coated with the resist is emitted from a light source having a wide wavelength width or a light source that effectively includes a plurality of wavelengths, and the spectral intensity of the light source is obtained. Irradiate light with different intensities so as to have a desired intensity distribution according to the wavelength including the influence of the distribution, image the alignment pattern irradiated with the light , and transfer it to the substrate and the substrate. Perform alignment with the reticle on which the pattern is formed, transfer the pattern of the reticle to the aligned substrate, expose the pattern, develop the exposed substrate by transferring the pattern, develop the developed pattern The relative position with respect to the alignment pattern was detected, and the accuracy of the overexposure of the developed pattern and the alignment pattern was determined. Superposing exposure method characterized by adjusting any of the exposure apparatus that exposes by transferring the pattern of the device or the reticle is coated with a resist in case it worse than the reference value exposure accuracy preset. 4. The method according to claim 3 , wherein an offset amount of alignment between the substrate and the reticle in the exposure apparatus is adjusted based on a result of the analysis. A resist is applied to the surface of the substrate on which the alignment pattern is formed, light is applied to the substrate on which the resist is applied, and reflected light from the substrate to which the light is applied has a desired intensity according to the wavelength of the reflected light. The alignment pattern is imaged in a state where the intensities are varied so as to be distributed, and the image obtained by the imaging is processed to align the substrate with the reticle on which the pattern to be transferred onto the substrate is formed. Transferring the pattern of the reticle onto the aligned substrate and exposing, developing the exposed substrate by transferring the pattern and detecting the relative position between the developed pattern and the alignment pattern. Obtain the accuracy of the overlaid exposure between the developed pattern and the alignment pattern, and if the obtained overlaid exposure accuracy is lower than a preset reference value, remove the resist. Superposing exposure method characterized by adjusting any of the exposure apparatus that to exposure transfer the pattern of the fabric to apparatus or the reticle. The overlay exposure method according to claim 5, wherein an offset amount of alignment between the substrate and the reticle in the exposure apparatus is adjusted based on a result of the analysis .

Images