JP3685667B2 - Three-dimensional shape detection method and apparatus, and inspection method and apparatus - Google Patents

Description

なお、ＮＡ’，ＮＡｓ’は夫々像側の開口数である。即ち、レンズの瞳にスリットを挿入してその開口を細長い長方形にすると、焦点面からｚ'だけ離れた面上での像（即ち、焦点はずれの像）が細長い長方形にぼけることがわかる。このとき、開口は矩形であることが重要なのではなく、一方向に大きく、それと直角方向に小さいことが重要なのである。
【００１７】
一方、開口の焦点面（ｘｙ平面）での像について、波動光学上、次式が成り立つ。

ここで、λは光の波長である。焦点面における像は、焦点はずれの像とは逆に、開口の大きい方向（ｙ方向）で短かくなり、図２（ｂ）のようになることがわかる。
【００１８】
ここで、長さｒｓと長さｒｓｏがほぼ等しくなるように、波長λや開口数ＮＡｓなどを選ぶと、ｘ方向のぼけ量はほぼ一定であって、ｙ方向のぼけ量を焦点はずれ量に比例して大きくすることができる。即ち、ｙ方向のぼけ量を何らかの手段で測定することにより、レンズと物体との間の距離を求めることができる。
【００１９】
これが共焦点方式の原理であり、上記の特開平４−２８３６０８号公報に記載の従来技術はこの原理を利用したものであるが、このように長さを測定するのではなく、図３（ａ）に示すように、レンズの焦点面に、図３（ｃ）に示すようなｘ方向に伸延してｙ方向をスリット幅とするスリットを配置し、このスリットを透過した光量を測定する方法もある。図３（ｂ）に示すようなｙ方向に細長い矩形開口のレンズの焦点はずれ像はｙ方向にぼけるので、この像をスリットを介して受光することにより、焦点はずれの度合い、従って、レンズから物体面までの距離に応じた光量が受光されることになる。ここで、このスリットの幅は、合焦点位置における結像した光像のｙ方向の幅と同程度にする。
【００２０】
この場合、このスリットの位置で像が合焦したときには、開口を通過した全光量がこのスリットを通過するが、焦点がはずれると、その光量の一部がスリットで遮られ、通過する光量が減少する。図３（ａ）に示すように、スリットの後に光センサを配置し、その受光量に応じた出力を測定する。センサ出力と物体の位置Ｚの関係は、図３（ｄ）に示すグラフのようになる。即ち、合焦位置（Ｚ＝０）でセンサの出力が最大となり、変位Ｚの絶対値が増加するととともにセンサ出力は減少する。
【００２１】
以上が本発明の原理であるが、その応用例として、図４に示すように、光路を分岐し、夫々の分岐光路毎にスリットと光センサとを配置し、一方の分岐光路では、これらスリットと光センサとがレンズの焦点位置の前に位置し、他方の分岐光路では、これらスリットと光センサとがレンズの焦点位置の後に位置するようにして、夫々の光センサの出力から物体の点０からの変位量Ｚを特定することができる。前ピンとなる光センサの出力をＡ、後ピンとなるセンサ出力をＢ、スリットなしのセンサの出力をＴとすると、物体の変位量Ｚは次式で与えられる。
Ｚ＝ｆ（Ａ／Ｔ，Ｂ／Ｔ）
ここで、Ｔは光路の分岐後の全光量に対するセンサの出力であり、各焦点位置における正規化光量がＡ／Ｔ，Ｂ／Ｔである。物体の変位量Ｚは、この正規化光量の関数として与えられる。
【００２２】
関数ｆは、変位量Ｚに対する各焦点位置におけるぼけを幾何光学的に求め、スリットを透過する割合を計算することによって得ることも可能であるが、高さ（即ち、変位量Ｚ）が既知であるテストサンプルを使用した実験でもって求めることも可能である。
【００２３】
また、図５に示すように、ｎ個（但し、ｎは２以上の整数）の光路を分岐し、夫々の光路毎にレンズの焦点位置の前後に複数のスリットと光センサを配置することにより、同様に、物体の変位量Ｚを特定することができる。この場合、ｎ個のスリット付センサを配置し、各センサの中で最大の出力を得てセンサの位置と結像関係にある共役位置を物体の位置とする。さらに、各センサの出力で得られるグラフを形成し、このグラフを補間処理することによって最大値の検出精度を高めることができる。
【００２４】
以上は物体上のある一点に着目していたが、以下、ライン上の各点の高さを同時に検出する方法について説明する。
【００２５】
図２では、矩形開口により、１点の結像が焦点はずれでｙ軸方向のみに像が広がることを説明した。ラインを点の集合と考えると、図６に示すように、ｘ軸上のラインの各点の像は、焦点はずれ量に応じて、ｙ軸方向に広がる。
【００２６】
このため、図７に示すように、光センサとしてラインセンサを使用すると、ラインセンサの画素の数だけ同時に光量を測定できる。矩形開口の形状や波長を選択することにより、ｘ方向のぼけ量をラインセンサの画素ピッチと同等になるように設定すると、ラインセンサの各検出画素（セル）は物体の異なる位置（これを、以下、単位領域という）からの光を検出することになる。従って、ライン上の各点の高さを同時に検出可能となる。さらに、物体をｙ軸と平行に移動させることにより、２次元領域の各点の高さを表わす画像（高さ画像）を得ることができる。なお、矩形開口は、瞳位置あるいは瞳の結像位置に空間フィルタを挿入することで容易に実現できる。
【００２７】
ラインセンサの配置には、図４に示した３つの光センサのように配置する方法と、図５に示したｎ個の光センサのように配置する方法とがある。これらラインセンサの互いに対応する位置の検出画素（セル）は、物体のこれに対応する位置の単位領域の焦点はずれ像を検出するようにする。このため、光学系はテレセントリック系が適切である。ラインセンサの前にはスリットが必要であるが、ラインセンサの開口をスリットとして代用することもできる。
【００２８】
次に、照明について考察する。
【００２９】
ここでは、ラインのみを照明するスリット状の照明、即ち、シートビームが必要である。照明光（スリット光）の幅は、ラインセンサ上で検出画素のサイズと同程度とする。照明は、物体の高さ測定範囲に相当するだけの焦点深度を有することが望ましい。焦点深度を大きくするために、照明の開口数ＮＡｂを小さくする。しかし、あまり小さくすると、回折の影響によってスリット光の幅を絞ることができない。このため、開口数ＮＡｂは、スリット光の幅が検出画素のサイズと同等になるように、次式のように設定すると良い。
ＮＡｂ≒λ／Ｇ
ここで、λは光の波長、Ｇは検出画素のサイズである。
【００３０】
本発明のポイントは、物体の複数の単位領域をスリット光で照明し、ｘ軸方向とｙ軸方向とで開口数が異なる開口絞りを配置することにより、ラインセンサの検出画素の配列方向の像のぼけ量を、焦点はずれ量によらず、ぼぼ一定にし、ラインセンサの画素画素の配列方向に直交する方向の像のぼけ量を焦点はずれ量に応じて変化するようにし、焦点面の前後に複数のラインセンサを配置し、各ラインセンサの各検出画素の出力から物体のスリット光で照射される線上の各単位領域の高さを算出するものにある。さらに、物体をステージに搭載し、ラインセンサの検出画素の配列方向とは直交する方向に移動させることにより、物体の表面の各点の高さ情報、即ち、立体形状を得るものである。
【００３１】
かかる構成によると、ラインセンサを用いているため、検出速度が速く、また、垂直落射照明、即ち、垂直上方から検出であるため、検査の死角がない。
【００３２】
上記立体形状検出系と物体の明るさを検出する平面画像検出系とを組み合わせて、検査対象物の明るさ画像と高さ画像とを検出し、パターン検査を行なう際、立体形状検出系の検出画素サイズを平面画像検出系の検出画素サイズの整数倍の大きさにすることにより、より高速な検査が可能となる。高さ画像の平面方向の解像度が明るさ画像に対して低いが、明るさ画像のみで検査する従来方法に比べて、検査速度は低下せずに高さの検査ができるメリットは大きい。
【００３３】
【発明の実施の形態】
以下、本発明の実施形態を図面により説明する。
図１に本発明による立体形状検出方法及び装置の第１の実施形態を示す構成図であって、１は検査対象物、２はＸＹステージ、３はドライバ、４はパーソナルコンピュータ、５はモータコントロール基板、６は画像入力基板、７ａ，７ｂ，７ｃはラインセンサ、８ａ，８ｂ，８ｃはアンプ、９ａ，９ｂ，９ｃはＡ／Ｄ変換回路、１０はレーザ源、１１はレンズ、１２はシリンドリカルレンズ、１３はコリメータレンズ、１４は穴付ミラー、１５は対物レンズ、１６は開口絞り、１７は結像レンズ、１８ａ，１８ｂはビームスプリッタである。
【００３４】
同図において、検査対象物１は、ＸＹステージ２上に搭載されてワークホルダを介して固定されている。照明光源としてのレーザ源１０は、例えば、半導体レーザであって、特に、発光点が直線形状のものが好適であり、例えば、１．５μｍ（半値幅）×２００μｍといった直線形状の物がある。勿論、これに限らず、気体レーザ源でもよい。
【００３５】
レーザ源１０から出射した光はレンズ１１とシリンドリカルレンズ１２によって直線状の光ビーム（スリット光）に形成され、コリメータレンズ１３と対物レンズ１５によって検査対象物１上を照射する。ここで、コリメータレンズ１３と対物レンズ１５との間には、この照射光を通過させる穴１４ａが設けられて検査対象物１からの反射光を反射させる穴付ミラー１４が設けられている。照明の開口数ＮＡｂはこの穴１４ａの径で規定することができる。勿論、開口数ＮＡｂは、レンズ１１を通過後のビーム径を適切な径に設定すること、あるいは、光路上適切な位置に絞り板（ピンホール）を挿入することで規定できるが、この穴付きミラー１４の穴１４ａを絞り板として兼用することもできる。また、穴付ミラー１４の代わりに、ハーフミラーを使用することも当然考えられる。
【００３６】
以上により、検査対象物１を、幅が検出画素サイズと同等で、長さが検出幅（＝ラインセンサの長さ／検出倍率）以上あるスリット光で照明する。
【００３７】
検査対象物１からの反射光は、検出光として、検出光路中のタンデム光学系をなす対物レンズ１５と結像レンズ１７とでＣＣＤのラインセンサ７ｃ上に結像する。対物レンズ１５と結像レンズ１７の光路中にある穴付きミラー１４は、このように、照明光路と検出光路を分離するものである。穴付きミラー１４の中心に開けた穴１４ａは、照明光を通過させる作用とともに、検出光の正反射光を遮断する作用がある。
【００３８】
対物レンズ１５と結像レンズ１７の光路中にある開口絞り１６は検出系の開口形状を規定するものであり、ｙ軸方向の開口数ＮＡに対してｘ軸方向の開口数ＮＡｓを小さくする。開口絞り１６の形状は、必ずしも矩形（即ち、スリット状）である必要はなく、角を丸めた矩形あるいは楕円や長円でもよい。この開口絞り１６により、ラインセンサ７ａ〜７ｃの検出画素（セル）の配列方向（図示するように配置されるラインセンサ７ｃでは、ｘ軸方向、ラインセンサ７ａ，７ｂでは、ｙ軸方向であるが、要するに、検出光路の光軸から見て、開口絞り１６の幅方向）のぼけ量を、像の焦点はずれ量によらず、ぼぼ一定にし、ラインセンサの検出画素の配列方向とは直交する方向（検出光路の光軸から見て、開口絞り１６の長手方向）のぼけ量を、像の焦点はずれ量に応じて変化するようにするものである。
【００３９】
結像レンズ１７を通過した検出光は、ビームスプリッタ１８ａ，１８ｂにより、その光路が３つに分岐される。夫々の分岐光路で光量が３等分されるようにしてもよいが、アンバランスがあっても構わない。できれば、ラインセンサ７ａ，７ｂでの光量がほぼ等しいことが望ましい。
【００４０】
分岐光路には夫々、ラインセンサ７ａ，７ｂ，７ｃ（以下では、これらラインセンサを総称して、ラインセンサ７という）が配置される。ラインセンサ７ａは高さ検出範囲の下限の物体面（検査対象物１の面）と結像関係にある像面位置に配置され、ラインセンサ７ｂは高さ検出範囲の上限の物体面と結像関係にある像面位置に配置され、ラインセンサ７ｃは、これら２つの像面位置の中間位置に配置されている。
【００４１】
なお、これらラインセンサ７は同一検出画素数、同一検出画素ピッチのセンサである。いま、検査対象物１のスリット光が照射される領域でのこれらラインセンサの検出画素に対応する微少領域を単位領域とすると、これらラインセンサ７のｎ番目の検出画素は、この検査対象物１のスリット光が照射される領域でのｎ番目の単位領域からの反射光を受光するように、これらラインセンサ７の位置が調整されている。これらラインセンサ７への駆動には、互いに同期した同一のクロックが供給される。ラインセンサ７ａ，７ｂとしては、検出画素が正方形、あるいは正方形に近いものが使用される。例えば、１４μｍ×１４μｍ（ピッチ×アパーチャ）のものがある。ラインセンサ７ｃとしては、検出画素が縦長のもの、例えば、１４μｍ×２００μｍ（ピッチ×アパーチャ）のものが使用される。
【００４２】
照明光、即ち、スリット光の幅は、ラインセンサ７の検出画素サイズとほぼ同じ大きさであるが、ラインセンサ７ａ，７ｂは、高さ検出範囲の上限と下限に相当する像面位置に配置されているため、検査対象物１からの反射光が常に焦点はずれの像として検出され、ラインセンサ７ａ，７ｂにおける検査対象物１の単位領域の像はこの検出画素より大きい。従って、ラインセンサ７ａ，７ｂの検出画素自体の大きさがスリットして作用し、ぼけた像の一部（中心部のみ）の光量を検出することになる。一方、ラインセンサ７ｃの検出画素は縦長であるため、焦点はずれのぼけた像全体の光量を検出できる。なお、ラインセンサ７ａ，７ｂの前にスリットを設けたり、ラインセンサ７ｃの前にシリンドリカルレンズを設けることにより、上記と同じ効果を得ることもできる。
【００４３】
ラインセンサ７ａ，７ｂ，７ｃの出力は夫々、アンプ８ａ，８ｂ，８ｃで増幅された後、Ａ／Ｄ変換器９ａ，９ｂ，９ｃでデジタル信号に変換され、画像入力基板２２から計算機、例えば、パーソナルコンピュータ（以下、パソコンという）４に供給される。ここで、ｍ個（但し、ｍは検査対象物１のスリット光が照射される領域での単位領域の個数に等しい）の検出画素を備えたラインセンサ７ａ，７ｂ，７ｃのｋ番目（但し、ｋ＝１，２，……，ｍ）の検出画素の出力値を夫々Ａ（ｋ），Ｂ（ｋ），Ｃ（ｋ）とすると、これら検出画素の結像関係にある検査対象物１での単位領域の高さＺ（ｋ）は、次式で与えられる。
Ｚ（ｋ）＝ｆ（Ａ（ｋ）／Ｃ（ｋ），Ｂ（ｋ）／Ｃ（ｋ））
ここで、関数ｆは、予め既知の高さを持つ標準サンプルを用いて測定され、テーブル形式でパソコン４に記憶されている。
【００４４】
また、変数Ａ（ｋ）／Ｃ（ｋ）は、高さ検出範囲の下限の物体面と結像関係にある像面位置における正規化した光量であり、変数Ｂ（ｋ）／Ｃ（ｋ）は、同じく上限における正規化した光量である。全光量で正規化することにより、検査対象物１の反射率の違いによる高さの計測誤差を低減できる。例えば、Ａ（ｋ）／Ｃ（ｋ）＝Ｂ（ｋ）／Ｃ（ｋ）であれば、高さＺ（ｋ）は高さ検出範囲のちょうど中間となる。また、Ａ（ｋ）／Ｃ（ｋ）＞Ｂ（ｋ）／Ｃ（ｋ）であれば、高さＺ（ｋ）は高さ検出範囲の下限に近い値となり、さらに、Ａ（ｋ）／Ｃ（ｋ）＜Ｂ（ｋ）／Ｃ（ｋ）であれば、高さＺ（ｋ）は高さ検出範囲の上限に近い値となる。この実施形態では、高さ検出範囲の下限と上限の物体面と結像関係にある像面位置にラインセンサを配置したが、ラインセンサの数を増やし、高さ検出範囲の途中に相当する面にもラインセンサを配置することにより、高さの検査精度の向上を図ることもできる。
【００４５】
なお、出力値Ｃ（ｋ）は、検査対象物１からの反射光の全光量であるから、明るさ画像に相当する。従って、この実施形態においては、高さ画像と明るさ画像とを同時に得ることができる。
【００４６】
パソコン４は、モータコントロール基板５からパルスを発生し、ドライバ３を駆動してＸＹステージ２を制御する。ラインセンサ７の検出画素の配列方向に直交する方向、即ち、検査対象物１に照射されるスリット光の幅方向（図示のｙ軸方向）にＸＹステージ２を一定速度で移動させながら、光量を検出することにより、スリット光で照射される帯状の範囲の各単位領域の高さを計測する。ＸＹステージ２をｙ軸方向に移動させてスリット光の照射領域が検査対象物の検査領域の端部に達すると、このＸＹステージ２をＸ方向に照射されるスリット光の長さ分検査対象物１を変位させ、次いで、ｙ軸方向のこれまでとは反対方向に検査対象物１を移動させるようにして、検査対象物１をｙ軸方向に往復移動させて、これにより、検査対象物１の検査領域全面の高さを単位領域毎に計測する。パソコン４及びモータコントロール基板５が管理しているＸＹステージ２の位置情報と、ラインセンサ７の出力から得られた高さ情報とから、検査対象物１の３次元の位置情報が得られる。
【００４７】
この第１の実施形態では、同じレンズ１５を用いて、検査対象物１にスリット光を垂直上方から照明し、検査対象物１からの反射光を垂直上方から検出するようにしている。このため、検査ができない死角の問題が発生しない。即ち、検査対象物１に対して同一方向からスリット光を照明・検出することが、この実施形態の１つの特徴である。また、図１で図示していないが、光学系全体を斜めに傾けて斜め方向からスリット光を検査対象物１に照明し、同じ斜め方向から反射光を検出する構成も実現できる。このとき、スリット光の幅方向に光学系を傾斜させる。
【００４８】
次に、この実施形態での光学系の諸元の一具体例を示す。
【００４９】
ｘ軸方向の検出開口数ＮＡｓと照明開口数ＮＡｂとはほぼ等しい値であって、ｙ軸方向の検出開口数ＮＡは、その５〜１０倍程度の値とするとよい。一例としては、

とする。
【００５０】
勿論、諸元は上記の２例に限るものではなく、また、ラインセンサ７の検出画素サイズは数μｍ〜数ｍｍの範囲であって、ｙ軸方向の検出開口数ＮＡはおよそ０．１〜０．９の範囲で光学系を構成できる。
【００５１】
また、ラインセンサ７としては、例えば、検出画素数２５６〜５０００個程度のＣＣＤリニアセンサを使用する。データレートは数ＭＨｚ〜２５ＭＨｚ程度で検出する。マルチタップタイプのセンサを使用すると、さらに高速な検出も可能となり、１００ＭＨｚを超えることも可能である。波長については、可視光に限定されるものではなく、赤外線や紫外線でもよい。
【００５２】
図８は本発明による立体形状検出方法及び装置の第２の実施形態を示す構成図であって、７ｄはラインセンサ、８ｄはアンプ、９ｄはＡ／Ｄ変換器、１８ｃはビームスプリッタ、１９はランプ、２０ａ，２０ｂはコンデンサレンズ、２１は拡散板、２２はスリット、２３は全反射ミラーであり、図１に対応する部分には同一符号を付けて重複する説明を省略する。
【００５３】
同図において、照明の光源としてランプ１９が使用される。このランプ１９は高輝度なキセノンランプや超高圧水銀ランプが好適であり、あるいは、ハロゲンランプを使用してもよい。ランプ１９から出射した光はコンデンサレンズ２０ａ，２０ｂによって集光され、その集光位置にスリット２２が配置される。コンデンサレンズ２０ａ，２０ｂで集光される光は、拡散板２１で照明むらが除かれるように拡散された後、スリット２２に送られてスリット光が形成される。また、図示していないが、ランプ１９の背面にコールドミラーを配置することにより、スリット光の光量を増加させることもできる。
【００５４】
スリット２２で形成されたスリット光は、コリメータレンズ１３と対物レンズ１５とにより、検査対象物１上を照射する。ここで、スリット２２と検査対象物１は結像関係にあり、また、スリット光は、その幅が上記の単位領域の幅に等しく（即ち、この単位領域の像がラインセンサ７ａ〜７ｄ（これらを総称して、ラインセンサ７という）の検出画素（セル）に投影されたとき、この像の幅がこの検出画素の配列方向の幅にほぼ等しい）、その長さは、ラインセンサ７ａ〜７ｄの検出画素数に等しい個数の単位領域の配列長さにほぼ等しい。ここで、ラインセンサ１９の検出画素配列方向の長さＬ_Lと検査対象物１でのスリット光の長さＬ_Sとの関係は、対物レンズ１５から結像レンズ１７までの検出系の検出倍率Ｇ₁に対し、
Ｌ_S＝Ｌ_L／Ｇ₁
の関係がある。
【００５５】
図１に示した第１の実施形態は、レーザ源１０を使用しているため、スペックル（干渉による明るさむら）が発生する。しかし、本発明では、像を検出しているのではなく、光量を検出しているため、スペックルの影響は少ないが、結像倍率などの光学系の条件によっては、スペックルのため、高さの検出精度が低下することが考えられる。図８に示した第２の実施形態では、光源としてランプ１９を用い、拡散板２１で光を拡散しているので、明るさのむらをなくすことができる。しかし、輝度の点では、レーザ源が勝っているので、場合によって使い分けるとよい。
【００５６】
次に、この第２の実施形態の検出系について説明する。
【００５７】
結像レンズ１７を通過した光は、ビームスプリッタ１８ａ，１８ｂ，１８ｃにより、４つの光路に分岐される。ここでは、分岐光路夫々の光量は等しいものとする。分岐光路毎にラインセンサ７ａ，７ｂ，７ｃ，７ｄが配置されている。なお、ラインセンサ７ｄの前に全反射ミラー２３が配置されているが、これは必ずしも必要なものではない。
【００５８】
これらラインセンサ７ａ，７ｂ，７ｃ，７ｄは夫々、異なる像面位置に配置されている。即ち、ラインセンサ７ａは高さ検出範囲の下限近くの物体面（検査対象物１の面）と結像関係にある像面位置に配置され、ラインセンサ７ｂはラインセンサ７ａよりも高さ検出範囲の下限，上限の中間と結像関係にある像面位置側に配置され、ラインセンサ７ｃはこの中間の像面位置よりも高さ検出範囲の上限側に配置され、ラインセンサ７ｄは高さ検出範囲の上限近くの物体面と結像関係にある像面位置に配置されている。
【００５９】
なお、これら４つのラインセンサ７は同一検出画素数、同一検出画素ピッチのセンサであって、夫々のｎ番目の検出画素は、物体面上のスリット光で照射される単位領域の像を検出するように、その位置が調整されている。また、これら４つのラインセンサ７の駆動クロックは、互いに同期した同一のクロックであり、検出画素の形状は正方形あるいは正方形に近いものであり、例えば、１３μｍ×１３μｍ（ピッチ×アパーチャ）のものがある。この場合、ラインセンサ７の検出画素自体の大きさがスリットして作用し、ぼけた像の一部（中心部のみ）の光量を検出することになる。
【００６０】
ラインセンサ７ａ，７ｂ，７ｃ，７ｄの検出出力は夫々、アンプ８ａ，８ｂ，８ｃ，８ｄで増幅された後、Ａ／Ｄ変換器９ａ，９ｂ，９ｃ，９ｄでデジタル信号に変換され、画像入力基板６により、パソコン４に入力される。ｍ個の検出画素のラインセンサ７ａ，７ｂ，７ｃ，７ｄのｋ番目の検出画素の出力値を夫々Ａ（ｋ），Ｂ（ｋ），Ｃ（ｋ），Ｄ（ｋ）とすると、図９に示すように、各検出画素（ｋ＝１，２，……，ｍ）の結像関係にある検査対象物１の高さＺ（ｋ）は、夫々の出力値Ａ（ｋ），Ｂ（ｋ），Ｃ（ｋ），Ｄ（ｋ）の間を補間し、これによって得られるグラフの最大となる点における高さとする。例えば、これら４つの出力値の最大値（図９では、Ｂ（ｋ））とその前後に位置する値（図９では、Ａ（ｋ），Ｃ（ｋ））との３つの出力値をガウス分布で補間し、その最大となる位置を求める。
【００６１】
この第２の実施形態では、４つのラインセンサ７を配置したが、ラインセンサの数をさらに増やすことにより、高さの検出精度が向上し、また、高さ検出範囲を拡大することができる。勿論、３個以下のラインセンサを用いるようにしてもよい。
【００６２】
なお、図１に示したレーザ源１０による照射系と図８に示した検出・処理系との組み合わせ、あるいは、図８に示したランプ１９による照射系と図１に示した検出・処理系との組み合わせも可能であることは勿論である。
【００６３】
図１０は本発明による立体形状検出方法及び装置の第３の実施形態を示す構成図であって、２４はポリゴンミラー、２５はリレーレンズであり、図１に対応する部分には同一符号を付けて重複する説明を省略する。
【００６４】
同図において、レーザ源１０から出射された光ビームは、ポリゴンミラー２４で反射された後、リレーレンズ２５と穴付きミラー１４の穴１４ａとを通過し、対物レンズ１５を介して検査対象物１に照射される。ここで、検査対象物１を照射する光ビームのスポットは、ポリゴンミラー２４の回転とともに、この検査対象物１の面を直線状に走査する。このスポットの径は、ラインセンサ７でのその像がラインセンサ７の検出画素の配列方向の幅とほぼ等しくなるように設定されており、このスポットの走査範囲は、ラインセンサ７の検出画素配列方向の長さをＬ_L、対物レンズ１５と結像レンズ１７との間の検出系の検出倍率をＧ₂とすると、
Ｌ_L／Ｇ₂
の関係にある。
【００６５】
このような関係からして、光ビームのスポットでこのように走査することは、先の第１，第２の実施形態でのスリット光を検査対象物１に照射することと同等である。
【００６６】
なお、この場合、ラインセンサ７ａ，７ｂ，７ｃの駆動をポリゴンミラー２４の回転と同期させるとよい。
【００６７】
また、ポリゴンミラーの代わりに、ガルバノミラーや偏向光学素子など他の手段を用いるようにしてもよい。
【００６８】
図１１は前出の立体形状検査方法及び装置を適用した本発明によるパターン検査方法及び装置の一実施形態を示す構成図であって、２ａはワークホルダ、６ａ，６ｂは画像入力ボード、６ｃはインターフェース基板（ＣＯＭ−Ｉ／Ｆ）、２６は光源、２７は光学系、２８はＣＣＤラインセンサ部、２９はＴＤＩラインセンサ、３０はローダ／アンローダ（Ｌ／ＵＬ）、３１はシーケンサ、３２はカメラコントロール・座標管理基板であり、前出図面に対応する部分には同一符号を付けている。
【００６９】
同図において、検査対象物１は、ＸＹステージ２にワークホルダ２ａを介して固定される。光学系２７は、図１や図８に示した高さ画像を得る立体形状検出系と明るさ画像を得る平面画像検出系とを複合した光学系である。立体形状検出系の検出器は３個のＣＣＤラインセンサを用いたＣＣＤラインセンサ部２８であり、図１に示されるアンプ８ａ〜８ｃやＡ／Ｄ変換器９ａ〜９ｃなども備えている。これら３個のＣＣＤラインセンサは、先に図１で説明した高さ検出範囲の下限の物体面と結像関係にある像面位置における明るさを検出するラインセンサと、高さ検出範囲の上限の物体面と結像関係にある像面位置における明るさを検出するラインセンサと、全体の明るさを検出するラインセンサである。
【００７０】
夫々ＣＣＤラインセンサから出力されてＡ／Ｄ変換された信号は、バスを介し、図１の画像入力基板６をなす画像入力ボード６ａからパソコン４に入力される。平面画像検出系の検出器は、マルチタップのＴＤＩ（Time Delay & Integration）ラインセンサ２９である。このＴＤＩセンサ２９は、ＣＣＤラインセンサに比べて高感度であるため、画素サイズを小さくしても高速な検出ができる。４タップのＴＤＩラインセンサを用い、検出信号のデータレートを４倍にすると、明るさ画像の検出画素のサイズを高さ画像の検出画素のサイズの半分にして、データ量が４倍になっても問題ない。例えば、高さ画像の検出画素サイズを６μｍ、明るさ画像の検出画素サイズを３μｍとし、ＣＣＤラインセンサ部２８のデータレートを１０ＭＨｚ、ＴＤＩラインセンサ２９のデータレートを１０ＭＨｚ×４＝４０ＭＨｚとする。ＴＤＩラインセンサ２９の出力信号は、図１の画像入力基板６をなす画像入力ボード６ｂを経由してパソコン４に入力される。
【００７１】
ＣＣＤラインセンサ部２８とＴＤＩラインセンサ２９とには、カメラコントロール・座標管理基板３２から駆動パルスが供給される。ＸＹステージ２のサーボモータに内蔵されているロータリエンコーダの出力パルス信号をドライバ３から得て、検出画素のサイズに相当するクロックに分周し、これを駆動パルスとしてＣＣＤラインセンサ部２８とＴＤＩラインセンサ２９とに供給する。ＣＣＤラインセンサ部２８には、ＸＹステージ２が６μｍ移動する毎に１パルス送り、ＴＤＩラインセンサ２９にも、ＸＹステージ２が６μｍ移動する毎に１パルス送る。これにより、ＸＹステージ２の移動速度とＣＣＤラインセンサ部２８とＴＤＩラインセンサ２９の駆動とが同期する。カメラコントロール・座標管理基板３２は、インターフェース基板６ｃを介して、パソコン４により制御させる。
【００７２】
パソコン４は、ステージコントロール基板５からパルスを発生し、ドライバ３を駆動してＸＹステージ２を制御する。ＣＣＤラインセンサ部２８のラインセンサの検出画素の配列方向とは直交する方向、即ち、ｙ軸方向（図１）にＸＹステージ２を一定速度で移動させながら、画像を検出する。ＸＹステージ２を、先に説明したように、往復動作させて、ＸＹステージ２のｙ軸方向の折り返し時、このＸＹステージ２をｘ軸方向に検出幅だけステップ送りすることにより、検査対象物１の検査対象領域のパターンを検出することができる。
【００７３】
ワークホルダ２ａや光源２６，ローダ／アンローダ３０は、シーケンサ３１によって制御される。ワークホルダ２ａやローダ／アンローダ３０はメカニカルな動作が制御対象であり、光源２６はシャッタの開閉や光量が制御対象である。シーケンサ３１はＲＳ−２３２Ｃでパソコン４と通信し、パソコン４が全体シーケンスを制御する。
【００７４】
パソコン４は、３つのＣＣＤラインセンサを備えたＣＣＤラインセンサ部２８の出力信号から、先に説明したように、高さ画像を生成するとともに、ＴＤＩラインセンサ２９の出力信号から明るさ画像を生成し、これらを用いてパターン欠陥を抽出する。高さ画像も用いているので、平面的な欠陥のみならず、厚み不足などの厚さ方向の欠陥をも検出できる。
【００７５】
以上のようにして得られた検査結果は、プリンタなどによって出力される。図１２はその出力結果の一例を示す図である。ここでは、検査結果としては、断線，ショート，かすれ，ピンホール，突起の各欠陥とする。ここでは、欠陥の種類毎に異なる形状のマークで、かつ検査の結果得られた位置情報に基づく位置に夫々表示する。従って、プリンタの出力紙に印字されたマークにより、夫々の欠陥の位置が示される。
【００７６】
また、かかる検査結果をネットワーク経由で検査装置外の計算機などへ転送することもできる。例えば、ＸＹステージと顕微鏡からなる欠陥確認ステーションに検査結果を転送し、その欠陥位置に基づいてＸＹステージを駆動させることにより、夫々の欠陥の確認作業を行なうことができる。このとき、欠陥の修正作業も同時に行なうことができる。
【００７７】
さらに、例えば、計算機に検査結果を保管しておき、統計処理などにより、製造装置のプロセスモニタとして活用することもできる。
【００７８】
図１３に先に説明した立体形状検出方法及び装置を適用した本発明によるはんだ付け部検査方法及び装置の一実施形態を示す構成図であって、図１１に対応する部分には同一符号を付けている。
【００７９】
同図において、検査対象物１は、ＸＹステージ２にワークホルダ２ａを介して固定される。光学系２７は、図１や図８に示した高さ画像を得る立体形状検出光学系である。立体形状検出系の検出器は３個のＣＣＤラインセンサとこれらの出力信号をＡ／Ｄ変換するＡ／Ｄ変換器などを含むＣＣＤラインセンサ部２８であり、これらＣＣＤラインセンサは、先に図１で説明した高さ検出範囲の下限の物体面と結像関係にある像面位置における明るさを検出するＣＣＤラインセンサと、高さ検出範囲の上限の物体面と結像関係にある像面位置における明るさを検出するＣＣＤラインセンサと、全体の明るさを検出するＣＣＤラインセンサである。夫々のＣＣＤラインセンサの出力信号は、Ａ／Ｄ変換された後、画像入力ボード６ａからパソコン４に入力される。例えば、ＣＣＤラインセンサの検出画素のサイズを１５μｍとし、ＣＣＤラインセンサのデータレートを１５ＭＨｚとする。
【００８０】
ＣＣＤラインセンサ部２８には、カメラコントロール・座標管理基板３２から駆動パルスが供給される。ＸＹステージ２のサーボモータに内蔵されているロータリエンコーダの出力パルス信号をドライバ３から得て、ＣＣＤラインセンサの検出画素のサイズに相当するクロックに分周し、駆動パルスとして、ＣＣＤラインセンサ部２８に供給する。ＣＣＤラインセンサ部２８には、ＸＹステージ２がｙ軸方向に１５μｍ移動する毎に１パルス送られる。これにより、ＸＹステージ２のｙ軸方向の速度とＣＣＤラインセンサの駆動とが同期される。カメラコントロール・座標管理基板３２は、インターフェース基板６ｃを介して、パソコン４で制御させる。
【００８１】
パソコン４は、ステージコントロール基板５からパルスを発生し、ドライバ３を駆動してＸＹステージ２を制御する。ＣＣＤラインセンサの検出画素の配列方向に直交する方向、即ち、ｙ軸方向（図１）にＸＹステージ２を一定速度で移動させながら、画像を検出する。先に説明したように、ＸＹステージ２をｙ軸方向で往復動作させ、ＸＹステージ２の折り返し時、ＸＹステージ２をｘ軸方向に検出幅だけステップ送りすることにより、検査対象物１の検査対象となるはんだ付け領域全体を検出する。
【００８２】
ワークホルダ２ａや光源２６，ローダ／アンローダ３０はシーケンサ３１で制御される。ワークホルダ２ａやローダ／アンローダ３０はメカニカルな動作が制御対象であり、光源２６はシャッタの開閉が制御対象である。シーケンサ３１は、ＲＳ−２３２Ｃでパソコン４と通信し、パソコン４が全体のシーケンスを制御する。
【００８３】
パソコン４は、ＣＣＤラインセンサ部２８の３個のＣＣＤラインセンサの出力信号から高さ画像を生成するとともに、全体の光量を検出しているＣＣＤラインセンサ（３個のＣＣＤラインセンサのうちの図１でのラインセンサ７ｃに相当するもの）の出力信号から明るさ画像を生成する。これら画像により、はんだ付部でのはんだ付の良否を判定し、はんだ付不良を検出する。
【００８４】
次に、本発明による検査装置の検査対象物について説明する。
【００８５】
１つの例は、セラミック基板に用いるグリーンシートに形成された金属微粒子の配線パターンである。欠陥として、断線や半断線，ショート，半ショート欠陥などのパターンの平面形状欠陥と、パターンの厚み方向の欠陥、例えば、かすれやピンホール（高さ不足系），突起（高さ過剰系）がある。
【００８６】
他の例は、プリント板に形成された銅配線パターンである。特に、ビルドアップ基板はパターン密度が高く、微細化しているため、小さな欠陥まで検出する必要がある。パターンのアスペクト比が大きくなり、パターン厚み方向の欠陥の検査も重要となる。欠陥として、断線や半断線，ショート，半ショート欠陥などのパターンの平面形状欠陥と、パターンの厚み方向の欠陥、例えば、薄いパターンやピンホール（高さ不足系），突起（高さ過剰系）がある。また、ビアホールやスルーホールの形状も検査対象となる。
【００８７】
さらに他の例は、フラットディスプレイの検査である。プラズマディスプレイパネル（ＰＤＰ）の隔壁高さや隔壁幅，蛍光体膜厚などの３次元形状検査や、隔壁製造工程における感光膜の検査に好適である。また、カラーフィルタの突起欠陥検査やブラウン管の蛍光体膜厚検査等にも適用できる。
【００８８】
さらに他の例は、プリント配線板のはんだ付検査である。ＱＦＰやＳＯＰ，チップ部品などのはんだ不良（濡れ不良，はんだ小，ブリッジなど）や位置ずれを検査する。
【００８９】
以上のように、電子回路関係のアプリケーションを例として挙げたが、他の分野への適用も勿論可能である。
【００９０】
【発明の効果】
以上説明したように、本発明によれば、検出器にラインセンサを使用しているため、機械的な可動部がなく、一定幅の通過領域を通過する光量に基づいて象のぼけ量を検出する方式をとり、ラインセンサの検出画素自体がこの通過領域としての機能をするので、別途スリットを設ける必要がなく、安価でかつ高さ検出を高速に行なうことができるし、検査対象物での光の照射方向と反射光の検出方向とを同じ方向とし、特に、垂直落射照明，垂直上方検出を行なうものであるから、検査の死角が生じない。
【００９１】
また、本発明による立体形状検出装置をパターン検査装置に使用することにより、高速で安価な検査装置が実現できる。
【００９２】
さらに、本発明による立体形状検出装置をはんだ検査装置に使用することにより、高速で安価な検査装置が実現できる。
【００９３】
さらに、本発明による立体形状検出装置を外観検査装置に使用することにより、高速で安価な検査装置が実現できる。
【図面の簡単な説明】
【図１】本発明による立体形状検出方法及び装置の第１の実施形態を示すブロック構成図である。
【図２】矩形開口レンズの焦点はずれ像を説明する図である。
【図３】焦点はずれ量を測定する原理を説明する図である。
【図４】焦点はずれ量を測定する第１の構成を説明する図である。
【図５】焦点はずれ量を測定する第２の構成を説明する図である。
【図６】矩形開口レンズによる線状物体の焦点はずれ像を説明する図である。
【図７】本発明の原理を説明する図である。
【図８】本発明による立体形状検出方法及び装置の第２の実施形態を示すブロック構成図である。
【図９】図１，図８に示した実施形態でのセンサ出力値よりを高さを算出する原理を説明する図である。
【図１０】本発明による立体形状検出方法及び装置の第３の実施形態を示すブロック構成図である。
【図１１】本発明によるパターン検査装置の一実施形態を示すブロック構成図である。
【図１２】図１１に示したパタン検査装置の検査結果の一例を示す図である。
【図１３】本発明によるはんだ検査装置の一実施形態を示すブロック構成図である。
【符号の説明】
１ 検査対象物
２ ＸＹステージ
３ ドライバ
４ 計算機
５ モータコントロール基板
６ 画像入力基板
７ａ〜７ｄ ラインセンサ
８ａ〜８ｄ アンプ
９ａ〜９ｄ Ａ／Ｄ変換器
１０ レーザ
１１ レンズ
１２ シリンドリカルレンズ
１３ コリメーティングレンズ
１４ 穴付ミラー
１４ａ 穴
１５ 対物レンズ
１６ 開口絞り
１７ 結像レンズ
１８ａ〜１８ｃ ビームスプリッタ
１９ ランプ
２０ａ，２０ｂ コンデンサレンズ
２１ 拡散板
２２ スリット
２３ 全反射ミラー
２４ ポリゴンミラー
２５ リレーレンズ[0001]
BACKGROUND OF THE INVENTION
The present invention relates to inspection of a three-dimensional shape of a predetermined object in appearance inspection of industrial products, and in particular, a wiring pattern of an electronic circuit board (for example, a wiring pattern printed on a printed wiring board or a ceramic green sheet) or an electronic circuit board. The present invention relates to an inspection of a suitable three-dimensional shape when inspecting the height of a soldered portion.
[0002]
[Prior art]
As a conventional example of inspection of a wiring pattern of metal fine particles formed on a green sheet used for a ceramic substrate, inspection methods disclosed in Japanese Patent Laid-Open Nos. 2-71377 and 6-27034 are known. In such a conventional inspection method, a wiring pattern on a green sheet is detected as a two-dimensional image by optical means, the detected image is used as a binarized image, and the connection relation of the wiring pattern represented by the binarized image is a design value. Are detected by detecting defects such as disconnection, semi-disconnection, short circuit, and semi-short circuit. As described above, this inspection method detects the wiring pattern as a two-dimensional image from the upper surface of the green sheet, and inspects the feature of the defect in the planar shape.
[0003]
On the other hand, it is also possible to detect a feature in the thickness direction of such a defect (that is, a thickness defect), for example, a fading of a wiring pattern, a pinhole (insufficient height), or a protrusion (excessive height). Various inspection methods for this purpose have been proposed.
[0004]
That is, for example, JP-A-3-279805, JP-A-4-290909, and JP-A-5-66118 disclose methods for inspecting wiring pattern thickness defects. This is a kind of light cutting method, which irradiates a wiring pattern with a light beam (laser light) and detects the reflected light. Either the irradiation direction of the light beam or the detection direction of the reflected light is used. One or both of them are inclined with respect to the substrate on which the wiring pattern is formed. The light beam scans the substrate on which this wiring pattern is formed in the x direction and sequentially scans in the y direction to scan the entire area of the substrate on which the wiring pattern is formed. In this case, the top surface is on a plane having the same height as the surface of the substrate. Therefore, in this case, even if the scanning position of the light beam changes, the reflected light to the sensor that receives the reflected light is reflected. The optical axis position does not change. On the other hand, when the thickness of the wiring pattern is different, the optical axis position of the reflected light with respect to the sensor changes in this portion. In this way, the optical axis of the reflected light with respect to the sensor differs according to the thickness (height) of the wiring pattern, and thereby the thickness of the wiring pattern can be detected.
[0005]
Conventional techniques for detecting the three-dimensional shape of an object are disclosed in, for example, Japanese Patent Application Laid-Open Nos. 3-63507 and 6-201337. This is a method of detecting a plurality of images having different focal positions while sequentially moving the Z stage on which the inspection target substrate is mounted in the height direction (vertical direction), and calculating a three-dimensional shape from these images. Japanese Patent Application Laid-Open No. 6-201337 also describes a method for calculating a three-dimensional shape by moving an optical system up and down.
[0006]
Another conventional technique for detecting the three-dimensional shape of an object is, for example, a method disclosed in Japanese Patent Laid-Open No. 5-240607. In this case, instead of moving the Z stage and the optical system, a sensor is provided corresponding to each position of the Z stage and the optical system. The target is irradiated with a light beam (laser light). In this method, reflected light is detected by sensors at a plurality of positions having different focal positions, and a three-dimensional shape is calculated from their signal intensities. In order to inspect the shape of the entire inspection area of the object, the inspection area is scanned with a light beam.
[0007]
As yet another conventional technique for detecting the three-dimensional shape of an object, there is a method disclosed in Japanese Patent Laid-Open No. 4-283608. This is called a confocal method, and when an image of an object is formed by a lens, if the surface of the object deviates from the focal point of the lens, the surface of the object at the focal plane of the lens This is based on the fact that the amount of blur varies depending on the amount of deviation of the surface of the object from the focal position of the lens. Specifically, the surface of the inspection object is irradiated with slit light, and the reflected light is detected by a two-dimensional area sensor through a slit-shaped aperture stop and a lens. The width is narrow enough to correspond to one pixel when light is received by the area sensor, and the length is long enough to correspond to multiple pixels of the area sensor (illuminated with this slit light). The small area that reflects the light of one pixel of the area sensor is hereinafter referred to as a unit irradiation area). The aperture stop has a narrow slit for one pixel received by the area sensor, and this slit is set perpendicular to the longitudinal direction of the slit light reflected from the object. Each pixel of the slit light reflected by the inspection object (reflected light from the unit irradiation area of the slit light of the inspection object) is received by different pixel areas of the area sensor via the aperture stop and the lens. However, the amount of blurring of the image at the area sensor of this unit irradiation region differs depending on the amount of deviation of the unit irradiation region from the focal point of the lens, and the larger this blur amount, the more the reflection from the unit irradiation region. The number of pixels that receive light increases. By detecting the number of pixels and performing calculations based on the number of pixels, the distance, that is, the height of the unit irradiation region from the lens can be obtained.
[0008]
[Problems to be solved by the invention]
By the way, since the three-dimensional shape detection method disclosed in the above Japanese Patent Laid-Open Nos. 3-279805, 4-290909, and 5-66118 uses a light beam, this light beam is used. In order to inspect the entire inspection area of the substrate to be inspected, it is necessary to scan the entire inspection area with a light beam. It is done using. However, in the case where the inspection is performed for each of the minute areas as described above, the inspection speed cannot be increased, and the inspection takes time. In addition, since the optical axis of the illumination light or reflected light is inclined with respect to the surface of the inspection target substrate, there is a problem that a blind spot that cannot be detected on the surface of the inspection target substrate occurs.
[0009]
The three-dimensional shape detection method described in the above Japanese Patent Laid-Open No. 3-63507 and Japanese Patent Laid-Open No. 6-201337 sequentially moves the Z stage and the optical system upward or downward, and obtains an image for each movement. In this way, since it is necessary to detect a plurality of images having different focal positions, there is also a problem that a long time is required for the inspection.
[0010]
Further, even in the three-dimensional shape detection method disclosed in the above Japanese Patent Laid-Open No. 5-240607, since the light beam is used, in order to inspect the entire inspection region of the inspection target substrate, the entire region is irradiated with the light beam. Therefore, the detection speed cannot be increased, and a long time is required for the inspection. Can't be fast. In addition, since a large number of sensors are required, these adjustments are complicated, and there is a problem in terms of cost.
[0011]
Furthermore, the three-dimensional shape detection method disclosed in the above-mentioned Japanese Patent Application Laid-Open No. 4-283608 can simultaneously detect the height of a plurality of unit irradiation areas, but blurs the image according to the distance from the lens to the object. Therefore, an expensive area sensor, i.e., an image sensor is required, and in this area sensor, the blur image is spread and received for each unit irradiation area. Therefore, it is necessary to determine the magnitude of the amount of received light for many pixels, which takes time for processing and takes a long time for inspection.
[0012]
The object of the present invention is to solve such a problem and to inspect the entire inspection area quickly and with high accuracy so that the defect in the inspection object can be inspected not only by the planar feature but also by the thickness. In addition, it is an object of the present invention to provide a three-dimensional shape detection method and apparatus capable of suppressing an increase in cost.
[0013]
[Means for Solving the Problems]
In order to achieve the above object, the present invention is based on a confocal method for detecting the height of an inspection object from the degree of blurring of the image of the inspection object due to defocusing. Used, each cell of the line sensor is made to correspond to a unit irradiation region of the inspection object, and the degree of image blur can be detected from the amount of light received by the cell.
[0014]
Here, the principle of the present invention will be described. First, with reference to FIG. 2, the defocusing in the case of a rectangular aperture (pupil is rectangular) will be considered.
[0015]
In FIG. 2A, the vertical direction on the paper is the y direction, the horizontal direction is the z direction, and the direction perpendicular to the paper is the x direction. When light emitted from the point 0 on the object plane in the z direction is passed through the lens, the image is formed on the focal plane. This focal plane varies depending on the distance from point 0 to the lens.
[0016]
Here, as shown in FIG. 2B, assuming that the numerical aperture of the long side of the rectangular aperture on the lens surface is NA, the numerical aperture of the short side is NAs, and the lateral magnification of the lens is β, 1 of the object plane An image on the xy plane separated by z ′ from the focal plane by the light emitted from the point 0 is substantially rectangular. If the length of the long side of this image is 2r and the length of the short side is 2rs, the following equation holds in terms of geometric optics.

NA 'and NAs' are numerical apertures on the image side. That is, when a slit is inserted into the pupil of the lens and its opening is made into a long and narrow rectangle, it can be seen that an image on a plane separated by z ′ from the focal plane (that is, an image out of focus) is blurred into a long and narrow rectangle. At this time, it is not important that the opening is rectangular, but it is important that the opening is large in one direction and small in the direction perpendicular thereto.
[0017]
On the other hand, the following equation is established in terms of wave optics for an image on the focal plane (xy plane) of the aperture.

Here, λ is the wavelength of light. It can be seen that the image on the focal plane becomes shorter in the large aperture direction (y direction), as shown in FIG.
[0018]
Here, when the wavelength λ, the numerical aperture NAs, etc. are selected so that the length rs and the length rso are substantially equal, the blur amount in the x direction is substantially constant, and the blur amount in the y direction is changed to the defocus amount. It can be increased proportionally. That is, the distance between the lens and the object can be obtained by measuring the amount of blur in the y direction by some means.
[0019]
This is the principle of the confocal method, and the conventional technique described in the above Japanese Patent Laid-Open No. 4-283608 utilizes this principle. However, instead of measuring the length in this way, FIG. As shown in FIG. 3, a method of measuring the amount of light transmitted through the slit by disposing a slit extending in the x direction and having a slit width in the y direction as shown in FIG. 3C on the focal plane of the lens. is there. The defocused image of the lens having a rectangular aperture elongated in the y direction as shown in FIG. 3B is blurred in the y direction. By receiving this image through the slit, the degree of defocusing, and therefore the object from the lens The amount of light corresponding to the distance to the surface is received. Here, the width of the slit is set to be approximately the same as the width in the y direction of the optical image formed at the in-focus position.
[0020]
In this case, when the image is focused at the position of this slit, the total amount of light that has passed through the aperture passes through this slit, but when the focus is lost, part of the amount of light is blocked by the slit, and the amount of light that passes is reduced. To do. As shown in FIG. 3A, an optical sensor is arranged after the slit, and an output corresponding to the amount of received light is measured. The relationship between the sensor output and the position Z of the object is as shown in the graph in FIG. That is, the sensor output becomes maximum at the in-focus position (Z = 0), and the absolute value of the displacement Z increases and the sensor output decreases.
[0021]
The above is the principle of the present invention. As an application example thereof, as shown in FIG. 4, the optical path is branched, and a slit and an optical sensor are arranged for each of the branched optical paths. And the optical sensor are positioned in front of the focal position of the lens, and in the other branch optical path, the slit and the optical sensor are positioned after the focal position of the lens, and the object point is determined from the output of each optical sensor. A displacement amount Z from 0 can be specified. Assuming that the output of the optical sensor serving as the front pin is A, the sensor output serving as the rear pin is B, and the output of the sensor without the slit is T, the displacement Z of the object is given by the following equation.
Z = f (A / T, B / T)
Here, T is the output of the sensor with respect to the total light quantity after branching of the optical path, and the normalized light quantity at each focal position is A / T and B / T. The displacement amount Z of the object is given as a function of this normalized light quantity.
[0022]
The function f can also be obtained by geometrically determining the blur at each focal position with respect to the displacement amount Z and calculating the ratio of transmission through the slit, but the height (ie, the displacement amount Z) is known. It can also be obtained by an experiment using a certain test sample.
[0023]
In addition, as shown in FIG. 5, n optical paths (where n is an integer of 2 or more) are branched, and a plurality of slits and optical sensors are arranged before and after the focal position of the lens for each optical path. Similarly, the displacement Z of the object can be specified. In this case, n sensors with slits are arranged, the maximum output among the sensors is obtained, and the conjugate position that is in an imaging relationship with the sensor position is set as the object position. Furthermore, by forming a graph obtained from the output of each sensor and performing interpolation processing on this graph, the detection accuracy of the maximum value can be increased.
[0024]
The above has focused on one point on the object. Hereinafter, a method for simultaneously detecting the height of each point on the line will be described.
[0025]
In FIG. 2, it has been explained that the image is spread out only in the y-axis direction because the image formation at one point is out of focus due to the rectangular aperture. Considering a line as a set of points, as shown in FIG. 6, the image of each point of the line on the x-axis spreads in the y-axis direction according to the defocus amount.
[0026]
For this reason, as shown in FIG. 7, when a line sensor is used as an optical sensor, the amount of light can be measured simultaneously by the number of pixels of the line sensor. When the amount of blur in the x direction is set to be equal to the pixel pitch of the line sensor by selecting the shape and wavelength of the rectangular aperture, each detection pixel (cell) of the line sensor has a different position of the object ( Hereinafter, light from a unit area) is detected. Therefore, the height of each point on the line can be detected simultaneously. Furthermore, an image (height image) representing the height of each point in the two-dimensional region can be obtained by moving the object parallel to the y-axis. Note that the rectangular opening can be easily realized by inserting a spatial filter at the pupil position or the pupil imaging position.
[0027]
The arrangement of the line sensor includes a method of arranging like the three photosensors shown in FIG. 4 and a method of arranging like the n photosensors shown in FIG. The detection pixels (cells) at the positions corresponding to each other of these line sensors detect an out-of-focus image of the unit region at the position corresponding to the object. For this reason, a telecentric system is appropriate for the optical system. A slit is required in front of the line sensor, but the opening of the line sensor can be used as a slit.
[0028]
Next, consider lighting.
[0029]
Here, slit-like illumination for illuminating only the line, that is, a sheet beam is necessary. The width of the illumination light (slit light) is approximately the same as the size of the detection pixel on the line sensor. It is desirable for the illumination to have a depth of focus that corresponds to the height measurement range of the object. In order to increase the depth of focus, the numerical aperture NAb of the illumination is decreased. However, if it is too small, the width of the slit light cannot be reduced due to the influence of diffraction. For this reason, the numerical aperture NAb is preferably set as in the following equation so that the width of the slit light is equal to the size of the detection pixel.
NAb ≒ λ / G
Here, λ is the wavelength of light, and G is the size of the detection pixel.
[0030]
The point of the present invention is to illuminate a plurality of unit regions of an object with slit light, and arrange an aperture stop having different numerical apertures in the x-axis direction and the y-axis direction, so that an image in the arrangement direction of detection pixels of the line sensor. The amount of blur is constant regardless of the amount of defocus, and the amount of image blur in the direction orthogonal to the pixel pixel array direction of the line sensor is changed according to the amount of defocus, before and after the focal plane. A plurality of line sensors are arranged, and the height of each unit region on the line irradiated with the slit light of the object is calculated from the output of each detection pixel of each line sensor. Furthermore, the height information of each point on the surface of the object, that is, a three-dimensional shape, is obtained by mounting the object on the stage and moving the object in a direction orthogonal to the arrangement direction of the detection pixels of the line sensor.
[0031]
According to such a configuration, since the line sensor is used, the detection speed is high, and since the vertical epi-illumination, that is, detection is performed from vertically above, there is no blind spot for inspection.
[0032]
The above three-dimensional shape detection system is combined with a planar image detection system that detects the brightness of an object to detect a brightness image and a height image of an inspection object and detect a three-dimensional shape detection system when performing pattern inspection. By making the pixel size an integer multiple of the detection pixel size of the planar image detection system, a higher-speed inspection can be performed. Although the resolution in the planar direction of the height image is lower than that of the brightness image, the advantage of being able to inspect the height without reducing the inspection speed is greater than that of the conventional method in which only the brightness image is inspected.
[0033]
DETAILED DESCRIPTION OF THE INVENTION
Embodiments of the present invention will be described below with reference to the drawings.
FIG. 1 is a block diagram showing a first embodiment of a three-dimensional shape detection method and apparatus according to the present invention, wherein 1 is an inspection object, 2 is an XY stage, 3 is a driver, 4 is a personal computer, and 5 is motor control. Substrate, 6 is an image input substrate, 7a, 7b and 7c are line sensors, 8a, 8b and 8c are amplifiers, 9a, 9b and 9c are A / D conversion circuits, 10 is a laser source, 11 is a lens, 12 is a cylindrical lens , 13 is a collimator lens, 14 is a mirror with a hole, 15 is an objective lens, 16 is an aperture stop, 17 is an imaging lens, and 18a and 18b are beam splitters.
[0034]
In the figure, an inspection object 1 is mounted on an XY stage 2 and fixed through a work holder. The laser source 10 as the illumination light source is, for example, a semiconductor laser, and in particular, a light emitting point having a linear shape is suitable, and for example, there is a linear shape of 1.5 μm (half width) × 200 μm. Of course, not limited to this, a gas laser source may be used.
[0035]
The light emitted from the laser source 10 is formed into a linear light beam (slit light) by the lens 11 and the cylindrical lens 12, and is irradiated on the inspection object 1 by the collimator lens 13 and the objective lens 15. Here, between the collimator lens 13 and the objective lens 15, a hole 14 a that allows the irradiation light to pass therethrough is provided, and a mirror 14 with a hole that reflects the reflected light from the inspection object 1 is provided. The numerical aperture NAb of illumination can be defined by the diameter of the hole 14a. Of course, the numerical aperture NAb can be defined by setting the beam diameter after passing through the lens 11 to an appropriate diameter, or by inserting a diaphragm plate (pinhole) at an appropriate position on the optical path. The hole 14a of the mirror 14 can also be used as a diaphragm plate. Of course, it is conceivable to use a half mirror instead of the holed mirror 14.
[0036]
As described above, the inspection object 1 is illuminated with the slit light whose width is equal to the detection pixel size and whose length is equal to or greater than the detection width (= line sensor length / detection magnification).
[0037]
Reflected light from the inspection object 1 is imaged on the CCD line sensor 7c as detection light by the objective lens 15 and the imaging lens 17 forming the tandem optical system in the detection optical path. The holed mirror 14 in the optical path of the objective lens 15 and the imaging lens 17 thus separates the illumination optical path and the detection optical path. The hole 14a formed in the center of the mirror 14 with a hole has an effect of blocking the regular reflection light of the detection light as well as an effect of passing the illumination light.
[0038]
An aperture stop 16 in the optical path between the objective lens 15 and the imaging lens 17 defines the aperture shape of the detection system, and reduces the numerical aperture NAs in the x-axis direction relative to the numerical aperture NA in the y-axis direction. The shape of the aperture stop 16 is not necessarily rectangular (that is, slit-shaped), and may be a rectangle with rounded corners, an ellipse, or an ellipse. By the aperture stop 16, the detection pixels (cells) of the line sensors 7a to 7c are arranged in the arrangement direction (in the x-axis direction in the line sensor 7c arranged as illustrated, and in the y-axis direction in the line sensors 7a and 7b). In short, when viewed from the optical axis of the detection optical path, the amount of blur in the width direction of the aperture stop 16 is made constant regardless of the amount of defocus of the image, and the direction orthogonal to the arrangement direction of the detection pixels of the line sensor. The blur amount in the longitudinal direction of the aperture stop 16 as viewed from the optical axis of the detection optical path is changed in accordance with the shift amount of the image.
[0039]
The detection light that has passed through the imaging lens 17 is split into three optical paths by beam splitters 18a and 18b. The amount of light may be divided into three equal parts in each branch optical path, but there may be unbalance. If possible, it is desirable that the light amounts at the line sensors 7a and 7b are substantially equal.
[0040]
Line sensors 7a, 7b, and 7c (hereinafter, these line sensors are collectively referred to as a line sensor 7) are arranged in the branched optical paths, respectively. The line sensor 7a is disposed at an image plane position having an imaging relationship with the lower limit object plane of the height detection range (the plane of the inspection object 1), and the line sensor 7b is imaged with the upper limit object plane of the height detection range. The line sensor 7c is arranged at an intermediate position between these two image plane positions.
[0041]
These line sensors 7 are sensors having the same number of detected pixels and the same detected pixel pitch. Now, assuming that a minute region corresponding to the detection pixel of these line sensors in the region irradiated with the slit light of the inspection object 1 is a unit region, the nth detection pixel of these line sensors 7 is the inspection object 1. The positions of these line sensors 7 are adjusted so as to receive the reflected light from the nth unit region in the region irradiated with the slit light. The same clocks synchronized with each other are supplied to drive the line sensors 7. As the line sensors 7a and 7b, detection pixels having a square shape or a shape close to a square shape are used. For example, there are 14 μm × 14 μm (pitch × aperture). As the line sensor 7c, one having a vertically long detection pixel, for example, 14 μm × 200 μm (pitch × aperture) is used.
[0042]
The width of the illumination light, that is, the slit light is substantially the same as the detection pixel size of the line sensor 7, but the line sensors 7a and 7b are arranged at image plane positions corresponding to the upper limit and the lower limit of the height detection range. Therefore, the reflected light from the inspection object 1 is always detected as an out-of-focus image, and the image of the unit area of the inspection object 1 in the line sensors 7a and 7b is larger than this detection pixel. Accordingly, the size of the detection pixels themselves of the line sensors 7a and 7b acts as a slit, and the amount of light of a part of the blurred image (only the central part) is detected. On the other hand, since the detection pixel of the line sensor 7c is vertically long, it is possible to detect the light amount of the entire image that is out of focus. The same effect as described above can be obtained by providing a slit in front of the line sensors 7a and 7b or providing a cylindrical lens in front of the line sensor 7c.
[0043]
The outputs of the line sensors 7a, 7b, and 7c are amplified by amplifiers 8a, 8b, and 8c, and then converted into digital signals by A / D converters 9a, 9b, and 9c. It is supplied to a personal computer (hereinafter referred to as a personal computer) 4. Here, the k-th (note that m is the line sensor 7a, 7b, 7c) provided with m detection pixels (where m is equal to the number of unit areas in the area irradiated with the slit light of the inspection object 1). Assuming that the output values of the detection pixels of k = 1, 2,..., m) are A (k), B (k), and C (k), respectively, the inspection object 1 is in the imaging relationship of these detection pixels. The unit region height Z (k) is given by the following equation.
Z (k) = f (A (k) / C (k), B (k) / C (k))
Here, the function f is measured in advance using a standard sample having a known height, and stored in the personal computer 4 in a table format.
[0044]
The variable A (k) / C (k) is a normalized light quantity at the image plane position that is in an imaging relationship with the lower limit object plane of the height detection range, and the variable B (k) / C (k). Is the normalized light intensity at the upper limit. By normalizing with the total amount of light, the height measurement error due to the difference in the reflectance of the inspection object 1 can be reduced. For example, if A (k) / C (k) = B (k) / C (k), the height Z (k) is exactly in the middle of the height detection range. If A (k) / C (k)> B (k) / C (k), the height Z (k) is close to the lower limit of the height detection range, and A (k) / If C (k) <B (k) / C (k), the height Z (k) is close to the upper limit of the height detection range. In this embodiment, line sensors are arranged at image plane positions that are in an imaging relationship with the lower limit and upper limit object planes of the height detection range, but the number of line sensors is increased, and the plane corresponding to the middle of the height detection range. In addition, it is possible to improve the inspection accuracy of the height by arranging the line sensor.
[0045]
The output value C (k) corresponds to a brightness image because it is the total amount of reflected light from the inspection object 1. Therefore, in this embodiment, a height image and a brightness image can be obtained simultaneously.
[0046]
The personal computer 4 generates a pulse from the motor control board 5 and drives the driver 3 to control the XY stage 2. While moving the XY stage 2 at a constant speed in the direction perpendicular to the arrangement direction of the detection pixels of the line sensor 7, that is, the width direction of the slit light irradiated to the inspection object 1 (the y-axis direction in the drawing) By detecting, the height of each unit area of the band-shaped range irradiated with the slit light is measured. When the XY stage 2 is moved in the y-axis direction and the irradiation area of the slit light reaches the end of the inspection area of the inspection object, the inspection object for the length of the slit light irradiated on the XY stage 2 in the X direction 1 is then displaced, and then the inspection object 1 is moved in the y-axis direction so as to move the inspection object 1 in the direction opposite to that in the y-axis direction. The height of the entire inspection area is measured for each unit area. From the position information of the XY stage 2 managed by the personal computer 4 and the motor control board 5 and the height information obtained from the output of the line sensor 7, the three-dimensional position information of the inspection object 1 is obtained.
[0047]
In the first embodiment, the same lens 15 is used to illuminate the inspection object 1 with slit light from above and detect reflected light from the inspection object 1 from above. For this reason, the problem of blind spots that cannot be inspected does not occur. That is, it is one of the features of this embodiment that the slit light is illuminated / detected from the same direction with respect to the inspection object 1. Although not shown in FIG. 1, it is also possible to realize a configuration in which the entire optical system is tilted obliquely and the inspection target 1 is illuminated with the slit light from an oblique direction, and the reflected light is detected from the same oblique direction. At this time, the optical system is inclined in the width direction of the slit light.
[0048]
Next, a specific example of the specifications of the optical system in this embodiment will be shown.
[0049]
The detected numerical aperture NAs in the x-axis direction and the illumination numerical aperture NAb are substantially equal, and the detected numerical aperture NA in the y-axis direction is preferably about 5 to 10 times that value. As an example,

And
[0050]
Of course, the specifications are not limited to the above two examples, and the detection pixel size of the line sensor 7 is in the range of several μm to several mm, and the detection numerical aperture NA in the y-axis direction is about 0.1 to 0.1 mm. An optical system can be configured in the range of 0.9.
[0051]
As the line sensor 7, for example, a CCD linear sensor having about 256 to 5000 detection pixels is used. The data rate is detected at several MHz to 25 MHz. If a multi-tap type sensor is used, even higher speed detection is possible, and it is possible to exceed 100 MHz. The wavelength is not limited to visible light, but may be infrared or ultraviolet.
[0052]
FIG. 8 is a block diagram showing a second embodiment of the method and apparatus for detecting a three-dimensional shape according to the present invention, in which 7d is a line sensor, 8d is an amplifier, 9d is an A / D converter, 18c is a beam splitter, 19 is Lamps, 20a and 20b are condenser lenses, 21 is a diffuser plate, 22 is a slit, and 23 is a total reflection mirror. The parts corresponding to those in FIG.
[0053]
In the figure, a lamp 19 is used as a light source for illumination. The lamp 19 is preferably a high-intensity xenon lamp or an ultra-high pressure mercury lamp, or a halogen lamp may be used. The light emitted from the lamp 19 is condensed by the condenser lenses 20a and 20b, and a slit 22 is disposed at the condensing position. The light condensed by the condenser lenses 20a and 20b is diffused by the diffusion plate 21 so that uneven illumination is removed, and then sent to the slit 22 to form slit light. Although not shown, the amount of slit light can be increased by arranging a cold mirror on the back of the lamp 19.
[0054]
The slit light formed by the slit 22 irradiates the inspection object 1 by the collimator lens 13 and the objective lens 15. Here, the slit 22 and the inspection object 1 are in an imaging relationship, and the slit light has a width equal to the width of the unit region (that is, the image of the unit region is represented by the line sensors 7a to 7d (these The width of the image is approximately equal to the width in the arrangement direction of the detection pixels when projected onto the detection pixels (cells) of the line sensor 7), and the length thereof is determined by the line sensors 7a to 7d. The number of unit regions equal to the number of detected pixels is approximately equal to the array length. Here, the length L of the line sensor 19 in the detection pixel arrangement direction_LAnd the length L of the slit light at the inspection object 1_SThe detection magnification G of the detection system from the objective lens 15 to the imaging lens 17 is₁Whereas
L_S= L_L/ G₁
There is a relationship.
[0055]
Since the first embodiment shown in FIG. 1 uses the laser source 10, speckle (brightness unevenness due to interference) occurs. However, in the present invention, since the amount of light is detected instead of detecting an image, the influence of speckle is small, but depending on the conditions of the optical system such as the imaging magnification, it is speckle. It is conceivable that the accuracy of detection will decrease. In the second embodiment shown in FIG. 8, since the lamp 19 is used as the light source and the light is diffused by the diffusion plate 21, uneven brightness can be eliminated. However, in terms of luminance, the laser source is superior, so it may be used properly depending on the case.
[0056]
Next, the detection system of the second embodiment will be described.
[0057]
The light that has passed through the imaging lens 17 is branched into four optical paths by beam splitters 18a, 18b, and 18c. Here, it is assumed that the light amounts of the branched light paths are equal. Line sensors 7a, 7b, 7c, and 7d are arranged for each branch optical path. In addition, although the total reflection mirror 23 is arrange | positioned in front of the line sensor 7d, this is not necessarily required.
[0058]
These line sensors 7a, 7b, 7c and 7d are arranged at different image plane positions. That is, the line sensor 7a is disposed at an image plane position that is in an imaging relationship with the object plane (the plane of the inspection object 1) near the lower limit of the height detection range, and the line sensor 7b is higher than the line sensor 7a. The line sensor 7c is arranged on the upper limit side of the height detection range from the intermediate image plane position, and the line sensor 7d detects the height. It is arranged at an image plane position that has an imaging relationship with the object plane near the upper limit of the range.
[0059]
These four line sensors 7 have the same number of detection pixels and the same detection pixel pitch, and each n-th detection pixel detects an image of a unit region irradiated with slit light on the object plane. So that its position is adjusted. The drive clocks of these four line sensors 7 are the same clocks synchronized with each other, and the shape of the detection pixel is a square or a square, for example, 13 μm × 13 μm (pitch × aperture). . In this case, the size of the detection pixel itself of the line sensor 7 acts as a slit, and the amount of light of a part of the blurred image (only the central part) is detected.
[0060]
The detection outputs of the line sensors 7a, 7b, 7c, and 7d are amplified by the amplifiers 8a, 8b, 8c, and 8d, respectively, are converted into digital signals by the A / D converters 9a, 9b, 9c, and 9d, and input to the image. The data is input to the personal computer 4 by the substrate 6. When the output values of the kth detection pixels of the line sensors 7a, 7b, 7c, and 7d of m detection pixels are respectively A (k), B (k), C (k), and D (k), FIG. As shown, the height Z (k) of the inspection object 1 in the imaging relationship of each detection pixel (k = 1, 2,..., M) is the output value A (k), B ( k), C (k), and D (k) are interpolated to obtain the height at the maximum point of the resulting graph. For example, three output values, the maximum value of these four output values (B (k) in FIG. 9) and the values located before and after (A (k), C (k) in FIG. 9), are expressed as Gauss. Interpolate with the distribution to find the maximum position.
[0061]
In the second embodiment, four line sensors 7 are arranged. However, by further increasing the number of line sensors, height detection accuracy can be improved and the height detection range can be expanded. Of course, three or less line sensors may be used.
[0062]
1 is combined with the detection / processing system shown in FIG. 8, or the irradiation system using the lamp 19 shown in FIG. 8 and the detection / processing system shown in FIG. Of course, a combination of these is also possible.
[0063]
FIG. 10 is a block diagram showing a third embodiment of the three-dimensional shape detection method and apparatus according to the present invention, in which 24 is a polygon mirror, 25 is a relay lens, and parts corresponding to those in FIG. Redundant description will be omitted.
[0064]
In the drawing, the light beam emitted from the laser source 10 is reflected by the polygon mirror 24, passes through the relay lens 25 and the hole 14 a of the holed mirror 14, and passes through the objective lens 15 to be inspected 1. Is irradiated. Here, the spot of the light beam that irradiates the inspection object 1 scans the surface of the inspection object 1 linearly as the polygon mirror 24 rotates. The diameter of the spot is set so that the image of the line sensor 7 is substantially equal to the width of the detection direction of the detection pixels of the line sensor 7, and the scanning range of the spot is the detection pixel array of the line sensor 7. Length in direction L_LThe detection magnification of the detection system between the objective lens 15 and the imaging lens 17 is set to G₂Then,
L_L/ G₂
Are in a relationship.
[0065]
From such a relationship, scanning in this manner with a light beam spot is equivalent to irradiating the inspection object 1 with the slit light in the first and second embodiments.
[0066]
In this case, the driving of the line sensors 7a, 7b, and 7c may be synchronized with the rotation of the polygon mirror 24.
[0067]
Further, other means such as a galvanometer mirror or a deflection optical element may be used instead of the polygon mirror.
[0068]
FIG. 11 is a block diagram showing an embodiment of a pattern inspection method and apparatus according to the present invention to which the above-described three-dimensional shape inspection method and apparatus are applied. 2a is a work holder, 6a and 6b are image input boards, and 6c is an image input board. Interface board (COM-I / F), 26 a light source, 27 an optical system, 28 a CCD line sensor unit, 29 a TDI line sensor, 30 a loader / unloader (L / UL), 31 a sequencer, 32 a camera This is a control / coordinate management board, and parts corresponding to the above-mentioned drawings are given the same reference numerals.
[0069]
In the figure, an inspection object 1 is fixed to an XY stage 2 via a work holder 2a. The optical system 27 is an optical system that combines the three-dimensional shape detection system that obtains the height image shown in FIGS. 1 and 8 and the planar image detection system that obtains the brightness image. The detector of the three-dimensional shape detection system is a CCD line sensor unit 28 using three CCD line sensors, and includes the amplifiers 8a to 8c and A / D converters 9a to 9c shown in FIG. These three CCD line sensors include a line sensor that detects brightness at an image plane position that has an imaging relationship with the lower limit object plane of the height detection range described above with reference to FIG. 1, and an upper limit of the height detection range. A line sensor that detects brightness at an image plane position that is in an imaging relationship with the object plane, and a line sensor that detects overall brightness.
[0070]
Signals output from the CCD line sensors and A / D converted are input to the personal computer 4 from the image input board 6a forming the image input board 6 of FIG. 1 via the bus. The detector of the planar image detection system is a multi-tap TDI (Time Delay & Integration) line sensor 29. Since the TDI sensor 29 is more sensitive than the CCD line sensor, high-speed detection can be performed even if the pixel size is reduced. If a 4-tap TDI line sensor is used and the data rate of the detection signal is quadrupled, the size of the detection pixel of the brightness image is half the size of the detection pixel of the height image, and the amount of data is quadrupled. There is no problem. For example, the detection pixel size of the height image is 6 μm, the detection pixel size of the brightness image is 3 μm, the data rate of the CCD line sensor unit 28 is 10 MHz, and the data rate of the TDI line sensor 29 is 10 MHz × 4 = 40 MHz. The output signal of the TDI line sensor 29 is input to the personal computer 4 via the image input board 6b that forms the image input board 6 of FIG.
[0071]
Drive pulses are supplied from the camera control / coordinate management board 32 to the CCD line sensor unit 28 and the TDI line sensor 29. The output pulse signal of the rotary encoder built in the servo motor of the XY stage 2 is obtained from the driver 3 and is divided into a clock corresponding to the size of the detected pixel, and this is used as a drive pulse for the CCD line sensor unit 28 and the TDI line. The sensor 29 is supplied. One pulse is sent to the CCD line sensor unit 28 every time the XY stage 2 moves 6 μm, and one pulse is sent to the TDI line sensor 29 every time the XY stage 2 moves 6 μm. As a result, the movement speed of the XY stage 2 and the drive of the CCD line sensor unit 28 and the TDI line sensor 29 are synchronized. The camera control / coordinate management board 32 is controlled by the personal computer 4 via the interface board 6c.
[0072]
The personal computer 4 generates a pulse from the stage control board 5 and drives the driver 3 to control the XY stage 2. The image is detected while moving the XY stage 2 at a constant speed in a direction orthogonal to the arrangement direction of the detection pixels of the line sensor of the CCD line sensor unit 28, that is, in the y-axis direction (FIG. 1). As described above, the XY stage 2 is reciprocated, and when the XY stage 2 is folded back in the y-axis direction, the XY stage 2 is stepped by a detection width in the x-axis direction to thereby inspect the inspection object 1. The pattern of the inspection target area can be detected.
[0073]
The work holder 2a, the light source 26, and the loader / unloader 30 are controlled by a sequencer 31. The work holder 2a and the loader / unloader 30 are controlled by a mechanical operation, and the light source 26 is controlled by opening / closing of a shutter and the amount of light. The sequencer 31 communicates with the personal computer 4 via RS-232C, and the personal computer 4 controls the entire sequence.
[0074]
The personal computer 4 generates a height image from the output signal of the CCD line sensor unit 28 having three CCD line sensors as described above, and also generates a brightness image from the output signal of the TDI line sensor 29. These are used to extract pattern defects. Since the height image is also used, not only planar defects but also defects in the thickness direction such as insufficient thickness can be detected.
[0075]
The inspection result obtained as described above is output by a printer or the like. FIG. 12 shows an example of the output result. Here, the inspection results are defects of disconnection, short circuit, fading, pinhole, and protrusion. Here, the mark is formed in a different shape for each type of defect and is displayed at a position based on position information obtained as a result of the inspection. Accordingly, the position of each defect is indicated by a mark printed on the output paper of the printer.
[0076]
Further, the inspection result can be transferred to a computer outside the inspection apparatus via the network. For example, each defect can be confirmed by transferring the inspection result to a defect confirmation station including an XY stage and a microscope and driving the XY stage based on the defect position. At this time, a defect correction operation can be performed simultaneously.
[0077]
Further, for example, the inspection result can be stored in a computer and used as a process monitor of a manufacturing apparatus by statistical processing or the like.
[0078]
FIG. 14 is a configuration diagram showing an embodiment of a soldering portion inspection method and apparatus according to the present invention to which the three-dimensional shape detection method and apparatus described above are applied, and the same reference numerals are given to the portions corresponding to FIG. ing.
[0079]
In the figure, an inspection object 1 is fixed to an XY stage 2 via a work holder 2a. The optical system 27 is a three-dimensional shape detection optical system that obtains the height image shown in FIGS. The detector of the three-dimensional shape detection system is a CCD line sensor unit 28 including three CCD line sensors and an A / D converter for A / D converting these output signals. These CCD line sensors are shown in FIG. CCD line sensor that detects brightness at an image plane position that is in an imaging relationship with the lower limit object plane of the height detection range described in 1, and an image plane that is in an imaging relationship with the upper limit object plane in the height detection range A CCD line sensor for detecting brightness at a position and a CCD line sensor for detecting overall brightness. The output signal of each CCD line sensor is A / D converted and then input to the personal computer 4 from the image input board 6a. For example, the size of the detection pixel of the CCD line sensor is 15 μm, and the data rate of the CCD line sensor is 15 MHz.
[0080]
A drive pulse is supplied from the camera control / coordinate management board 32 to the CCD line sensor unit 28. The output pulse signal of the rotary encoder built in the servo motor of the XY stage 2 is obtained from the driver 3 and is divided into a clock corresponding to the size of the detection pixel of the CCD line sensor, and the CCD line sensor unit 28 is used as a drive pulse. To supply. One pulse is sent to the CCD line sensor unit 28 every time the XY stage 2 moves 15 μm in the y-axis direction. Thereby, the speed of the XY stage 2 in the y-axis direction and the drive of the CCD line sensor are synchronized. The camera control / coordinate management board 32 is controlled by the personal computer 4 via the interface board 6c.
[0081]
The personal computer 4 generates a pulse from the stage control board 5 and drives the driver 3 to control the XY stage 2. An image is detected while moving the XY stage 2 at a constant speed in the direction orthogonal to the arrangement direction of the detection pixels of the CCD line sensor, that is, in the y-axis direction (FIG. 1). As described above, the inspection object 1 is inspected by reciprocating the XY stage 2 in the y-axis direction and step-feeding the XY stage 2 by the detection width in the x-axis direction when the XY stage 2 is folded back. The entire soldering area is detected.
[0082]
The work holder 2a, the light source 26, and the loader / unloader 30 are controlled by a sequencer 31. The work holder 2a and the loader / unloader 30 are controlled by a mechanical operation, and the light source 26 is controlled by opening and closing of a shutter. The sequencer 31 communicates with the personal computer 4 via RS-232C, and the personal computer 4 controls the entire sequence.
[0083]
The personal computer 4 generates a height image from the output signals of the three CCD line sensors of the CCD line sensor unit 28, and detects the total light amount (a diagram of the three CCD line sensors). 1 (corresponding to the line sensor 7c at 1) is generated from the output signal. From these images, the quality of soldering at the soldering part is determined, and a soldering failure is detected.
[0084]
Next, the inspection object of the inspection apparatus according to the present invention will be described.
[0085]
One example is a wiring pattern of metal fine particles formed on a green sheet used for a ceramic substrate. Defects include planar shape defects such as breaks, semi-breaks, shorts, and semi-short defects, and defects in the thickness direction of the pattern, such as blurring, pinholes (insufficient height system), and protrusions (excessive height system). is there.
[0086]
Another example is a copper wiring pattern formed on a printed board. In particular, since the build-up substrate has a high pattern density and is miniaturized, it is necessary to detect even a small defect. The pattern aspect ratio increases, and inspection of defects in the pattern thickness direction is also important. Defects include planar shape defects such as breaks, semi-breaks, shorts, and semi-short defects, and defects in the thickness direction of the pattern, such as thin patterns, pinholes (under-height), and protrusions (over-height). There is. In addition, the shape of via holes and through holes is also an inspection target.
[0087]
Yet another example is the inspection of flat displays. It is suitable for three-dimensional shape inspection such as partition wall height, partition wall width and phosphor film thickness of a plasma display panel (PDP), and inspection of a photosensitive film in a partition wall manufacturing process. Further, the present invention can be applied to color filter protrusion defect inspection, cathode ray tube phosphor film thickness inspection, and the like.
[0088]
Yet another example is a soldering inspection of a printed wiring board. QFP, SOP, chip parts and other solder defects (wetting defects, small solder, bridges, etc.) and displacement are inspected.
[0089]
As described above, the application relating to the electronic circuit is given as an example, but it is of course possible to apply to other fields.
[0090]
【The invention's effect】
As described above, according to the present invention, since a line sensor is used as a detector, there is no mechanical moving part, and the amount of blur of an elephant is detected based on the amount of light passing through a fixed width passing area. Since the detection pixel itself of the line sensor functions as this passing region, there is no need to provide a separate slit, and it is possible to perform high-speed detection at a low cost and at a high speed. Since the irradiation direction of light and the detection direction of reflected light are the same direction, and in particular, vertical epi-illumination and vertical upward detection are performed, no blind spot of inspection occurs.
[0091]
Further, by using the three-dimensional shape detection apparatus according to the present invention for a pattern inspection apparatus, a high-speed and inexpensive inspection apparatus can be realized.
[0092]
Furthermore, by using the three-dimensional shape detection apparatus according to the present invention for a solder inspection apparatus, a high-speed and inexpensive inspection apparatus can be realized.
[0093]
Furthermore, by using the three-dimensional shape detection apparatus according to the present invention for an appearance inspection apparatus, a high-speed and inexpensive inspection apparatus can be realized.
[Brief description of the drawings]
FIG. 1 is a block diagram showing a first embodiment of a solid shape detection method and apparatus according to the present invention.
FIG. 2 is a diagram illustrating an out-of-focus image of a rectangular aperture lens.
FIG. 3 is a diagram for explaining the principle of measuring a defocus amount.
FIG. 4 is a diagram illustrating a first configuration for measuring a defocus amount.
FIG. 5 is a diagram illustrating a second configuration for measuring a defocus amount.
FIG. 6 is a diagram for explaining a defocused image of a linear object by a rectangular aperture lens.
FIG. 7 is a diagram illustrating the principle of the present invention.
FIG. 8 is a block configuration diagram showing a second embodiment of a solid shape detection method and apparatus according to the present invention.
9 is a diagram for explaining the principle of calculating the height from the sensor output value in the embodiment shown in FIG. 1 and FIG.
FIG. 10 is a block diagram showing a third embodiment of a solid shape detection method and apparatus according to the present invention.
FIG. 11 is a block diagram showing an embodiment of a pattern inspection apparatus according to the present invention.
12 is a diagram showing an example of an inspection result of the pattern inspection apparatus shown in FIG.
FIG. 13 is a block configuration diagram showing an embodiment of a solder inspection apparatus according to the present invention.
[Explanation of symbols]
1 Inspection object
2 XY stage
3 Driver
4 Calculator
5 Motor control board
6 Image input board
7a-7d Line sensor
8a-8d amplifier
9a-9d A / D converter
10 Laser
11 Lens
12 Cylindrical lens
13 Collimating lens
14 Mirror with hole
14a hole
15 Objective lens
16 Aperture stop
17 Imaging lens
18a-18c Beam splitter
19 Lamp
20a, 20b condenser lens
21 Diffuser
22 Slit
23 Total reflection mirror
24 polygon mirror
25 Relay lens

Claims (15)

A plurality of unit areas on a straight line of the inspection object are illuminated, and reflected light from the plurality of unit areas illuminated by the inspection object is connected to the lens through a lens and a rectangular opening. Light is received by a plurality of line sensors arranged at different positions before and after the image plane, and the reflected light from the unit area that is different for each detection pixel is detected by each line sensor. The height of each unit region that is illuminated by the inspection object is determined by the amount of light detected according to the amount of blur of the image in the width direction of the light illumination region of the inspection object that is obtained for each unit region by the sensor. A three-dimensional shape detection method characterized by calculating. An illumination system for illuminating a plurality of unit areas on a straight line of an inspection object;
And a small shape of the opening and lens in a direction to increase and perpendicular thereto in one direction, an optical system for the reflected light from the plurality of the unit areas to pass,
A plurality of line sensors that are disposed at different positions before and after the imaging plane of the optical system and detect reflected light from the plurality of unit regions that have passed through the optical system ;
Processing means for calculating the height information of each of the plurality of unit areas illuminated by the detection outputs of the plurality of line sensors ,
Each of the plurality of line sensors detects a reflected light from a separate unit area illuminated by the light of the inspection object for each detection pixel . In claim 2,
A three-dimensional shape detection apparatus, wherein a width of a light illumination area of the inspection object in the line sensor is substantially the same as a size of the detection pixel of the line sensor. The plurality of unit areas on a straight line of the inspection object are illuminated with light, the inspection object is continuously moved in the width direction of the light illumination area of the inspection object , and the plurality of light illuminated on the inspection object The reflected light from the unit area is received by the line sensors arranged at different positions before and after the lens and the rectangular aperture, and detected by each of the line sensors. Detection of the reflected light from the unit area that differs for each pixel, and detection according to the amount of blur of the image in the width direction of the light illumination area of the inspection object obtained for each unit area by each of the line sensors A solid shape detection method , wherein the height of the unit area of the inspection object is calculated from the amount of light . Illuminating a plurality of unit areas on a straight line of the inspection object, and reflecting light reflected from the plurality of unit areas illuminated by the inspection object in the same direction as the illumination direction to the plurality of unit areas , Detected by a line sensor arranged at a plurality of different positions before and after the imaging plane of the lens via an optical system having a lens and a rectangular opening, and each line sensor separately detects each detected pixel. the resulting detection amount of the unit region of the by the detection amount of light, three-dimensional shape detection method characterized by calculating the height of the unit each area being light illumination of the inspection object. An illumination system for illuminating a plurality of unit areas on a straight line of an inspection object;
A stage for continuously moving the inspection object in the width direction of the light illumination area by the illumination system ;
An optical system having a lens through which reflected light from the plurality of unit areas illuminated by the inspection object passes and a rectangular aperture;
A plurality of line sensors for detecting the reflected light from the object to be examined through the optical system in the unit separate detection pixels for each region of the inspection object,
Processing means for calculating the height for each unit area from the amount of light detected for each detection pixel of the plurality of line sensors ,
The three-dimensional shape detection apparatus , wherein the plurality of line sensors are arranged at different positions before and after the imaging plane of the lens of the optical system . An illumination system for illuminating a plurality of unit areas on a straight line of an inspection object;
Of the reflected light from the plurality of unit areas illuminated by the inspection object, the reflected light in the same direction as the illumination direction to the inspection object by the illumination system is extracted, and the extracted reflected light passes through an optical system having an aperture of lenses and rectangular to,
A plurality of line sensors that are arranged at different positions before and after the imaging plane of the optical system and detect the reflected light through the optical system separately for each unit region that is illuminated by the light of the inspection object ; ,
A three-dimensional shape detection apparatus comprising: processing means for calculating a height for each unit region from a light amount for each unit region detected by each of the line sensors. In claim 6 or 7,
A three-dimensional shape detection apparatus, wherein a data rate of the processing means is several MHz to several tens of MHz. A plurality of unit regions on a straight line of the inspection object are illuminated, the inspection object is continuously moved in the width direction of the light illumination area of the inspection object, and the inspection object is illuminated. Reflected light from a plurality of unit regions is received by a plurality of line sensors arranged at different positions before and after the lens and a rectangular aperture, and each line sensor receives the light. The reflected light from the unit area that is different for each detection pixel is detected, and the line sensor according to the amount of image blur in the width direction of the light illumination area of the inspection object obtained for each unit area A pattern inspection method for detecting a pattern defect in the inspection object by calculating a height of the unit area illuminated by the inspection object using a detected light amount. An illumination system for illuminating a plurality of unit areas on a straight line of an inspection object;
A stage for continuously feeding the inspection object in the width direction of the light illumination area by the illumination system ;
The inspection object has a small aperture in the direction corresponding to the longitudinal direction of the light illumination area and a large shape in the direction corresponding to the width direction of the light illumination area, and the lens is illuminated. An optical system that allows the reflected light from the plurality of unit regions to pass through,
The plurality of unit regions which are respectively arranged at a plurality of different positions parallel to the optical axis of the reflected light passing through the optical system and before and after the imaging plane of the optical system , and are illuminated by the inspection object A plurality of line sensors having detection pixels for separately detecting the reflected light from each ;
Processing means for calculating the brightness image of the light- illuminated area of the inspection object and the height information for each unit area from the amount of light detected for each detection pixel of the plurality of line sensors;
A pattern inspection apparatus comprising: defect detection means for extracting a pattern defect on the inspection object from the brightness image obtained from the processing means and the height information. In claim 10,
Pattern inspection apparatus, wherein the size of the detection pixels of the height information is an integral multiple of the size of the detected pixel brightness image. A plurality of unit areas on a straight line of the soldering part of the inspection object are illuminated , and reflected light from the plurality of unit areas illuminated by the inspection object is passed through the lens and the rectangular opening, Light is received by a plurality of line sensors arranged at different positions before and after the imaging plane of the lens, and the reflected light from the unit area that is different for each detection pixel is detected by each of the line sensors. calculating the height of the each unit area from the detected amount of light corresponding to the blur amount of the image in the width direction of the light illumination area the unit said test object obtained for each region by the line sensor in the soldering portion A soldering inspection method characterized in that a height image is obtained and a soldering defect in the soldering portion is detected from the obtained height image. An illumination system for illuminating a plurality of unit areas on a straight line of a soldering portion of an inspection object;
A stage for continuously feeding the inspection object in the width direction of the light illumination area by the illumination system ;
The inspection object has a small aperture in the direction corresponding to the longitudinal direction of the light illumination area and a large shape in the direction corresponding to the width direction of the light illumination area, and the lens is illuminated. An optical system that allows the reflected light from the plurality of unit regions to pass through,
Before or after the image plane of the optical system, and are respectively arranged in a plurality of different positions parallel to the optical axis of the reflected light passing through the optical system, a unit of said plurality of being light illumination the test object A plurality of line sensors having detection pixels for separately detecting the reflected light from each region;
Processing means for calculating a brightness image of the light illumination area of the inspection object and height information for each unit area from the amount of light detected for each detection pixel of the plurality of line sensors;
A soldering inspection apparatus comprising: defect detection means for extracting a soldering defect in the inspection object from the brightness image obtained from the processing means and the height information. A plurality of unit areas on a straight line of an inspection object are illuminated , and reflected light from the plurality of unit areas illuminated by the inspection object is imaged through the lens and a rectangular opening. Light is received by a plurality of line sensors arranged at different positions before and after the surface, and the reflected light from the unit area that is different for each detection pixel is detected by each of the line sensors. The height information for each unit area of the inspection object that is light-illuminated is calculated from the detected light amount corresponding to the blur amount of the image in the width direction of the light illumination area of the inspection object that is obtained for each unit area. An appearance inspection method characterized by detecting a shape defect of the inspection object based on the height information. An illumination system for illuminating a plurality of unit areas on a straight line of an inspection object;
A stage for continuously feeding the inspection object in the width direction of the light illumination area by the illumination system ;
The inspection object has a small aperture in the direction corresponding to the longitudinal direction of the light illumination area and a large shape in the direction corresponding to the width direction of the light illumination area, and the lens is illuminated. An optical system that allows the reflected light from the plurality of unit regions to pass through,
The plurality of unit regions which are respectively arranged at a plurality of different positions parallel to the optical axis of the reflected light passing through the optical system and before and after the imaging plane of the optical system , and are illuminated by the inspection object A plurality of line sensors having detection pixels for separately detecting the reflected light from each ;
And processing means for calculating the height information of the brightness image and the unit for each area of the area light illuminated to the test object from the light amount detected for each detection pixel of the plurality of line sensors, said processing means An appearance inspection apparatus comprising: defect detection means for extracting a shape defect in the inspection object from the brightness image obtained from the information and the height information.

Images