JP3751660B2 - Regular pattern defect inspection system - Google Patents

Description

【０１０１】
のように示される。この場合の２次元空間フィルタ処理とは、ある点（ｉ，ｊ）を中心としたマスクサイズ３×３の領域の各点の濃度値を、重みＮ
【０１０２】
【数１】

【０１０３】
で演算して、新しい濃度ｆ（ｉ，ｊ）、即ち、
【０１０４】
【数２】

【０１０５】
を求める処理をいう。
【０１０６】
例４
2×(A4+B4)-(A1+B1)-(A7+B7），2×(A5+B5)-(A2+B2)-(A8+B8），2×(A6+B6)-(A3+B3)-(A9+B9），2×(B4+C4)-(B1+C1)-(B7+C7），2×(B5+C5)-(B2+C2)-(B8+C8），2×(B6+C6)-(B3+C3)-(B9+C9），…
これは、各パターンについて２画素の濃度値を総和を比較する他の例である。例えば、2×(A4+B4)-(A1+B1)-(A7+B7）の演算は、第２の規則的パターン１０２の左側部の２画素の総濃度を２倍し、これから第１及び第３の規則的パターン１０１及び１０３の同様の２画素の総濃度を減算するものである。なお、１次元のラインセンサにより取得したデータを用いて上記の演算を行う場合は、例３の場合と同様に、パイプライン処理機を空間フィルタ回路として用いると良い。
【０１０７】
本例の場合も、複数のライン（２ライン）にわたる比較演算を行っているため、規則的パターンの配列方向がラインセンサカメラ６０の走査方向に対して傾いている場合でも、好適に欠陥を検出することができる。
【０１０８】
例５
2×(A4+A5+B4+B5)-(A1+A2+B1+B2)-(A7+A8+B7+B8），2×(A5+A6+B5+B6)-(A2+A3+B2+B3)-(A8+A9+B8+B9），2×(A6+A7+B6+B7)-(A3+A4+B3+B4)-(A9+A10+B9+B10）
，… これは、各パターンについて４画素（２行２列）の濃度値の総和を比較するものである。例えば、2×(A4+A5+B4+B5)-(A1+A2+B1+B2)-(A7+A8+B7+B8）の演算は、第２の規則的パターン１０２を包含する４画素の総濃度を２倍し、これから第１及び第３の規則的パターン１０１及び１０３の同様の４画素の総濃度を減算するものである。この演算も、例４と同様に、パイプライン処理機を用いて行うことができる。
【０１０９】
例６
2×(A4+B4+C4)-(A1+B1+C1)-(A7+B7+C7)，2×(A5+B5+C5)-(A2+B2+C2)-(A8+B8+C8)，2×(A6+B6+C6)-(A3+B3+C3)-(A9+B9+C9)，…
これは、各パターンについて、３ラインにわたる３画素、すなわち被検査物１０の移動方向に沿った３画素の濃度値の総和を比較するものである。例えば、2×(A4+B4+C4)-(A1+B1+C1)-(A7+B7+C7)の演算は、第２の規則的パターン１０２の左側部の３画素の総濃度を２倍し、これから第１及び第３の規則的パターン１０１及び１０３の同様の３画素の総濃度を減算するものである。なお、１次元のラインセンサにより取得したデータを用いて上記の演算を行う場合は、図１３の欠陥検出部２８を用いると良い。また、本例の演算によっても、図１４のようなパターンの欠陥を好適に検出することができる。
【０１１０】
以上のように、画面上の同一形状の局所領域であって規則的パターンの繰り返し周期の整数倍の周期で配列された領域における画像濃度を比較することで、各領域の近傍に位置する規則的パターンを比較し、規則的パターンの欠陥を検出する。比較する各領域の幅は、相互の領域が重複しないように、領域の配列周期の画素数（本実施例では、３画素）以内とすると良い。
【０１１１】
本実施例の欠陥検出方法には、比較する局所領域を構成する画素数や比較する局所領域の幅を規定する画素数に応じて次のような特徴がある。
【０１１２】
例えば、局所領域が１画素の場合、部分欠陥の検出感度が良好であるという利点がある。ただし、この場合は、ノイズの影響を受けやすくノイズが欠陥と認識される誤検出が比較的発生しやすい。
【０１１３】
局所領域の幅を規定する画素数が中程度の数ｎ（１＜ｎ＜繰り返し周期の画素数）である場合は、複数の画素の濃度を加算して比較するため、ノイズとのレベル差が大きくなりＳ／Ｎが向上する。このため、部分的な欠陥であってもその検出感度を低下させることなく誤検出を抑制することができる。また、規則的パターンの配列方向とラインセンサカメラ６０の走査方向とに多少のずれがある場合にも、複数のラインで比較することにより好適に欠陥を検出することができる。ただし、局所領域の画素数を決定することはあまり容易でない。
【０１１４】
局所領域の幅を規定する画素数が規則的パターンの繰り返し周期の画素数（本実施例では、３画素）である場合も、複数画素の濃度を加算する結果、Ｓ／Ｎが高く誤検出が少ない。また、規則的パターンの配列方向とラインセンサカメラ６０の走査方向とに多少のずれがある場合にも、複数のラインで比較することで好適に欠陥を検出することができる。ただし、部分的な欠陥を検出するには多少不向きな面がある。
【０１１５】
なお、一般的には、比較する局所領域の構成画素数をｎとし、このｎ個の画素濃度を加算することでＳ／Ｎは√ｎ倍となる。これにより、ノイズの影響が低減され誤検出が抑制される。
【０１１６】
このように、局所領域の選択によって欠陥検査の特徴が異なることになる。したがって、規則的パターンの態様や欠陥の態様に応じて、また要求される欠陥検出の程度や精度、検出所要時間などに応じて、局所領域の幅や局所領域を構成する画素数を選択すると良い。
【０１１７】
次に、比較すべき局所領域の配列周期は規則的パターンの繰り返し周期を規定する画素数の整数倍であれば良いが、図１４のように規則的パターンの配列方向がラインセンサカメラ６０の走査方向に対して傾いている場合を考慮すると、局所領域の配列周期は小さい方が好ましい。これは、配列周期が大きい程、各局所領域と各パターンとの位置関係のずれが大きくなり、正確な比較が難しくなるためである。
【０１１８】
実施例２
図１５は、本実施例の欠陥検査装置の構成を示す概略図である。本実施例では、検査すべき規則的パターンを撮像する装置が、スキャナ８０及び長尺ＰＭＴ（Photomultiplier tube；光電子増倍管）８６から構成されている。
【０１１９】
スキャナ８０は、半導体レーザを光源とする検査光を照射するもので、１軸ステージ１２の上方に配置されている。このスキャナ８０は、規則的パターンの配列方向に沿って所定の速度で規則的パターンを走査する。また、スキャナ８０は、スキャナ・コントローラ８２を介して欠陥検出部２８に接続されており、このスキャナ・コントローラ８２によって駆動制御されている。
【０１２０】
長尺ＰＭＴ８６は、規則的パターンの配列方向に沿って延びる受光面を有するもので、スキャナ８０に対向して１軸ステージ１２の下方に設置されており、被検査物１０を透過したスキャナ８０からの検査光を受光するようになっている。
【０１２１】
長尺ＰＭＴ８６と１軸ステージ１２との間には、被検査物１０を透過した検査光を拡散する拡散板が設置されており、被検査物１０を透過した光にむらが生じないようにされている。
【０１２２】
また、長尺ＰＭＴ８６には、長尺ＰＭＴ８６の出力信号に対しＡ／Ｄ変換を行うＡ／Ｄ変換回路９０が接続されており、このＡ／Ｄ変換回路９０には欠陥検出部２８が接続されている。
【０１２３】
図１６は、本実施例の欠陥検出部２８の構成を示すブロック図である。この欠陥検出部２８は、長尺ＰＭＴ８６のシェーディング補正を行うシェーディング補正回路２８ａ、長尺ＰＭＴ８６の暗電流補正を行う暗電流補正回路２８ｂ、これらの補正のためにデータの加算処理を行う加算器２８ｃと、２次微分等の空間フィルタ処理を行う空間フィルタ回路２８ｄと、予め設定したレベルによる３値化を行うＬＵＴ２８ｅと、及びエッジ検出を行うエッジ検出回路２８ｆから構成されている。
【０１２４】
欠陥検出部２８は、マンマシンイターフェイスの機能をも有する制御部１６に接続されており、制御部１６により統括制御されるようになっている。この制御部１６には、図示はしないが、規則的パターンの欠陥を表示するモニタ等の表示装置、検出データを記録するＦＤ、ＨＤ、ＭＯ等の記録装置、検出データを受けとるホストコンピュータ等の外部機器がそれぞれ接続されている。
【０１２５】
次に、本実施例の欠陥検査装置の動作、及び本実施例の装置を用いた欠陥検査方法について説明する。
【０１２６】
まず、検査を開始するに先立って、検査条件の設定を行う。具体的には、実施例１と同様にして、長尺ＰＭＴ８６及びスキャナ８０からなる撮像装置のシェーディング補正データ、並びに長尺ＰＭＴ８６の暗電流補正データを取得する。シェーディング補正データを取得する際は、スキャナ８０を駆動させ、検査光を出射させておく。暗電流補正データを取得する際は、スキャナ８０を停止させ、検査光が出射しないようにしておく。
【０１２７】
また、検査前に被検査物１０を撮像し、１ラインずつ取り込んだデジタル画像信号に対してシェーディング補正回路２８ａ及び暗電流補正回路２８ｂにより撮像データの補正を行った後、空間フィルタ回路２８ｄをスルーにしてＬＵＴ２８ｅにより２値化を行ってから、エッジ検出回路２８ｆによりエッジ検出を行う。エッジ検出回路２８ｆにより検出された疑似欠陥に対応するエッジ位置データは、制御部１６に送出され、保存される。これにより、検査中は、制御部１６が疑似欠陥に対応するエッジ位置の間に現れる欠陥のみを有効とし、その範囲以外は比較演算の対象から除外するようになる。
【０１２８】
以上の設定作業が終了したら、検査を開始する。まず、搬送ステージ１２を駆動させておき、スキャナ８０を動作させる。これにより、搬送ステージ１２上を移動する被検査物１０に、スキャナ８０からの検査光が照射される。
【０１２９】
図１７に示されるように、被検査物１０を透過した光は、拡散板を通過して長尺ＰＭＴ８６に受光される。スキャナ８０が規則的パターンを１ラインスキャンする間、長尺ＰＭＴ８６がサンプリングを行うことで、１ラインの画像データが取得される。この画像データ信号は、Ａ／Ｄ変換回路９０でデジタル信号に変換された後、欠陥検出回路２８に転送される。
【０１３０】
１ラインずつ取り込んだデジタル画像信号は、シェーディング補正回路２８ａによってシェーディング補正が施され、暗電流補正回路２８ｂによって暗電流補正が施された後、加算器２８ｃをスルーして空間フィルタ回路２８ｄに入力される。空間フィルタ回路２８ｄは、実施例１と同様の空間フィルタ処理を施し、欠陥部を強調するとともにオフセットレベルを一定にする。空間フィルタ回路２８ｄから出力された画像データはＬＵＴ２８ｅに入力されて３値化され、この後、エッジ検出回路２８ｆに入力される。エッジ検出回路２８ｆは、この画像信号に基づいてエッジ検出を行い、そのデータを制御部１６に転送する。制御部１６は、エッジ検出回路２８ｆから転送されたデータに基づき、パターン欠陥の種類、位置、大きさを検出する。制御部１６は、これらの検出結果に基づき、必要に応じて、表示装置３０ａに規則的パターンの欠陥を表示したり、記録装置３０ｂに検出データを記録したり、外部機器３０ｃに検出データを出力したり、また記録・出力の際にデータを加工する。
【０１３１】
本実施例の欠陥検査装置も、３個の規則的パターンについてその近傍に位置する画素濃度を比較することで、欠陥の有無のみならず、欠陥の種類、位置、大きさ等を検出することができ、詳細な欠陥検査が可能である。また、比較の際、検査すべき規則的パターンについての画素濃度を２倍してから、その前後のパターンについての画素濃度を減算するので、これにより欠陥が強調され、Ｓ／Ｎが向上する。これにより、検査精度を高めることができる。
【０１３２】
さらに、本実施例の欠陥検査装置では、長尺ＰＭＴ８６というポイントセンサにより検査光を受光するので、ラインセンサを用いる場合に比べて干渉縞（モアレ）が発生しないという利点があり、高精度の欠陥検査が一層容易である。また、検査条件の設定がラインセンサを用いる場合に比べて容易であり、このため、効率の良い欠陥検査が可能である。
【０１３３】
【発明の効果】
本発明の欠陥検査装置によれば、所定の３個の領域の画像濃度Ｓ１、Ｓ２及びＳ３を比較し、これによって画像上の規則的パターンのうちこれらの領域の近傍に位置する第１、第２及び第３のパターンを比較するので、欠陥の有無のみならず、欠陥の種類、位置、大きさ等を容易に検出することできる。
【０１３４】
本発明の欠陥検査装置のうち演算により生じる疑似欠陥信号を除去して欠陥を検出するものによれば、疑似欠陥が真の欠陥であると誤認されることを防止されるので、検出精度の高い欠陥検査を行うことができる。
【０１３５】
本発明の欠陥検査装置のうち搬送ステージを備えるものによれば、規則的パターンを有する被検査物をこの搬送ステージによって継続的に移動させながら撮像を行うことで画像情報が順次に欠陥検出部に送出され、効率良く欠陥を検出することができる。
【０１３６】
本発明の欠陥検査装置において、撮像部が規則的パターンの配列方向に沿った１次元の画像情報を取得するものであると、撮像部の構成を簡略なものとすることができ、装置全体の小型化や製造コストの低減を達成することができる。
【０１３７】
欠陥検出部が複数ラインにわたる領域の画像濃度を比較する場合は、１次元の画像情報を取得する方向と規則的パターンの配列方向とが一致しない場合にも、各パターンを包含するようなライン数にわたる領域の画像濃度を比較するように設定することで、好適に欠陥を検出することができる。
【０１３８】
欠陥検出部がシェーディング補正を行う手段を備えていると、撮像に用いられる検査光の照射むらや検査光を受光する手段の感度むらに起因するシェーディングが画像情報に与える影響が低減され、誤検出を防止することができるので、検出精度の高い欠陥検査を行うことができる。
【０１３９】
同様に、欠陥検出部が暗電流補正を行う手段を備えていると、撮像部の暗電流が画像情報に与える影響が低減され、誤検出を防止することができるので、検出精度の高い欠陥検査を行うことができる。
【図面の簡単な説明】
【図１】実施例１に係る規則的パターンの欠陥検査装置の構成を示す概略図である。
【図２】被検査物１０上の規則的パターンの態様を示す模式図である。
【図３】受光手段２０の構成と制御部１６による制御を説明するための図である。
【図４】ラインセンサカメラ６０及び欠陥検出部２８の構成を示すブロック図である。
【図５】実施例１の欠陥検出部２８の回路図である。
【図６】規則的パターンの配列方向とラインセンサカメラ６０の走査方向との関係を示す図である。
【図７】各種の画像データ転送タイミング方式に関し、ラインセンサ系３２、制御系３４及び搬送系３６の関係を示すブロック図である。
【図８】規則的パターンと画素との関係を示す図である。被検査物の規則的パターンが傾いている場合の被検査物の規則的パターンとラインセンサカメラの画素との関係を説明するための模式図である。
【図９】空間フィルタ回路２８ｄのフィルタパターンを示す図である。
【図１０】規則的パターンに張り欠陥がある場合の欠陥検出部２８における演算結果を説明するための図である。
【図１１】規則的パターンに欠け欠陥がある場合の欠陥検出部２８における演算結果を説明するための図である。
【図１２】規則的パターンに各種の欠陥がある場合の欠陥検出部２８における演算結果を模式的に示す図である。
【図１３】パイプライン形の空間フィルタ回路２８ｄ′を示す回路図である。
【図１４】規則的パターンの配列方向がラインセンサカメラ６０の走査方向に対して傾いている様子を示す図である。
【図１５】実施例２の欠陥検査装置の構成を示す概略図である。
【図１６】本実施例２の欠陥検出部２８の構成を示すブロック図である。
【図１７】スキャナ８０と長尺ＰＭＴ８６との関係を示す模式図である。
【符号の説明】
１０…被検査物、１２…１軸ステージ、１４…ステージ・コントローラ、１６…制御部、１８…ライン光源、２０…受光手段、２８…欠陥検出部、２８ａ…シェーディング補正回路、２８ｂ…暗電流補正回路、２８ｃ…加算器、２８ｄ…空間フィルタ回路、２８ｅ…ＬＵＴ、２８ｆ…エッジ検出回路、３０ａ…表示装置、３０ｂ…記録装置、３０ｃ…外部機器、３２…ラインセンサ系、３４…制御系、３６…搬送系、４４…ステージ・コントローラ、４８…エンコーダ、５６…シャッタ、５８…レンズ、６０…ラインセンサカメラ、６０ａ…ラインセンサ素子、６０ｂ…アナログフィルタ回路、６０ｃ…Ａ／Ｄ変換回路、６１…回転ステージ、６２…レンズフォーカス・コントローラ、６４…高さ調整用ステージ、６６…ラインセンサ・コントローラ、６８…シェーディングデータメモリ、７０…ＡＬＵ、７１…ＡＬＵ、７２…暗電流データメモリ、７３…メモリ、７４…ＡＬＵ、７６ａ及び７６ｂ…１Ｈ遅延回路、７８…ＡＬＵ、８０…スキャナ、８２…スキャナ・コントローラ、８６…長尺ＰＭＴ、９０…Ａ／Ｄ変換回路。[0001]
[Industrial application fields]
The present invention relates to an apparatus for inspecting defects in regular patterns, in particular optical regular patterns.
[0002]
[Prior art]
Conventionally, as a defect inspection apparatus for inspecting defects in an optical regular pattern, the regular pattern obtained by imaging a regular pattern with a two-dimensional camera or a line sensor camera while transferring the regular pattern to be inspected is obtained. 2. Description of the Related Art There has been known an apparatus for inspecting a defect by comparing each regular pattern image after storing the pattern image in a memory.
[0003]
Further, a regular pattern to be inspected is imaged by a CCD (Charge Coupled Device) line sensor to obtain an image corresponding to the pattern, and the image corresponds to an integer multiple of one period of pixels (picture elements) of the regular pattern. Defect inspection apparatus (Japanese Patent Laid-Open No. 55-74409; “Repeated pattern defect inspection method” in which a pattern in each period is recognized by comparing with pixel information for each period of an image signal. And a regular pattern to be inspected is transferred in the pattern arrangement direction and imaged by a CCD line sensor to obtain a signal corresponding to the sum of the number of pixels for each repetition period. A defect inspection apparatus (Japanese Patent Laid-Open No. 55-87431; “A”) that determines the presence or absence of a defect by comparing a signal corresponding to the number of pixels with a signal corresponding to a standard number of pixels. "Repeated pattern defect inspection method" has been proposed.
[0004]
[Problems to be solved by the invention]
However, a defect inspection apparatus using a two-dimensional camera has a problem that inspection requires a long time because it requires two-stage processing of performing comparison processing after performing image accumulation processing. In addition, since regular pattern transfer must be temporarily stopped for imaging, a mechanism for that purpose is required, and in order to acquire data in two dimensions, devices such as a light source and a conveyor are large. Therefore, there is a problem that the entire apparatus is increased in size and cost.
[0005]
In addition, in a defect inspection apparatus using a line sensor camera, it cannot be said that a commercially available line sensor camera has a sufficiently high dynamic range, so that it is not sufficient as an apparatus for detecting a defect of a minute regular pattern. There was a problem.
[0006]
Also, in an apparatus using a CCD line sensor camera, since the object to be imaged is a regular pattern, interference fringes (moire) occur between the regular pattern and the CCD chip, thereby increasing the false positive rate of defects. There was a problem.
[0007]
In the above-mentioned “Repeated pattern defect inspection method” of JP-A-55-74409, since comparison is performed for each pixel, it is easily affected by noise mixed in the input system, and this noise is erroneously recognized as a defect. There was a problem that it was easy to do. In addition, there is a problem that a defect is easily recognized as a defect even when it is affected by interference fringes generated between the regular pattern and the CCD chip.
[0008]
When smoothing or the like is performed in order to reduce noise or the like in order to avoid such erroneous recognition, false recognition of pseudo defects is reduced, but defect recognition is likely to occur. Also, when the imaging time is lengthened to reduce noise, false recognition of pseudo defects is reduced, but the defect detection time is lengthened. In addition, since the inspection line speed must be slowed down in accordance with the imaging time, this cannot be realized when the line speed cannot be slowed down.
[0009]
In addition, in the “Repeated pattern defect inspection method” disclosed in Japanese Patent Laid-Open No. 55-87431, the sum of the number of pixels in a regular pattern is targeted, so that a positive defect such as a partial tension in the regular pattern is included. And a negative defect with a partial defect exist at the same time, there is a problem that it is difficult to detect the defect.
[0010]
The present invention has been made in view of the above circumstances, and an object of the present invention is to provide an apparatus capable of detecting a defect of a regular pattern in detail and with high accuracy.
[0011]
[Means for Solving the Problems]
In order to achieve the above object, a defect inspection apparatus for a regular pattern according to the present invention inspects defects of a plurality of regular patterns arranged at a predetermined repetition period along a predetermined direction on an inspection object. (A) an imaging unit that images the regular pattern and obtains image information of the regular pattern; and (b) based on the image information, a repetition period of the regular pattern on the image. The image density S1, S2 and S3 of the three regions arranged along the arrangement direction of the regular pattern with a period of an integral multiple is obtained, and the regular pattern is obtained by comparing these image densities S1, S2 and S3. A defect detection unit for detecting the defect.
[0012]
The defect detection unit may perform 2 × S2-S1-S3 calculation using the image densities S1, S2, and S3, and detect a defect of a regular pattern based on the calculation result. More preferably, the defect detection unit detects a defect by removing a pseudo defect signal generated by the above calculation.
[0013]
The defect inspection apparatus of the present invention may further include (c) a transfer stage that moves the inspection object along a predetermined direction. At this time, the conveyance stage may continuously move the inspection object, and the imaging unit may send the image information to the defect detection unit at a predetermined time interval. The conveyance stage may continuously move the inspection object, and the imaging unit may send the image information to the defect detection unit every time the inspection object moves a predetermined distance.
[0014]
The imaging unit included in the defect inspection apparatus of the present invention may acquire one-dimensional image information along a predetermined direction. At this time, the defect detection unit simultaneously processes the one-dimensional image information of a plurality of lines along the direction orthogonal to the arrangement direction of the regular pattern, so that the three regions described at the beginning and covering the plurality of lines. It is preferable to compare image densities of objects.
[0015]
As a first aspect of the imaging unit, (1) a line light source having a light emitting part extending in the arrangement direction of a regular pattern and irradiating the inspection light emitted from the light emitting part on the object to be inspected, {Circle around (2)} A one-dimensional image pickup device having a light receiving portion extending in a predetermined direction, which includes an inspection light that receives light that has passed through an inspection object, can be considered. At this time, it is preferable to further include means for adjusting the arrangement of the one-dimensional image sensor. Further, as a second mode, (1) a light source that irradiates the inspection object with inspection light and scans the inspection light along the arrangement direction of the regular pattern, and (2) the arrangement direction of the regular pattern. A light detection means having a light-receiving surface extending along the light-receiving surface and receiving light that has passed through the inspection object among inspection light is conceivable.
[0016]
The defect detection unit preferably includes means for performing shading correction on the image information, and detects defects based on the image information corrected thereby. In addition, the defect detection unit may include a unit that performs dark current correction on the image information and detects a defect based on the image information corrected thereby.
[0017]
[Action]
The defect inspection apparatus according to the present invention compares the image densities S1, S2 and S3 of the three regions arranged in the cycle as described above, so that the regular pattern on the image is positioned in the vicinity of these regions. The first, second and third patterns to be compared are compared. Defect detection is performed by this comparison. This comparison is achieved, for example, by performing an operation of 2 × S2-S1-S3. In this case, a defect of the second pattern is detected. By repeating the same comparison with respect to the image density of another three regions arranged in the same manner as the above region, it is possible to compare various three patterns arranged in the same cycle as those regions.
[0018]
By comparing the three patterns, it becomes easy to detect not only the presence / absence of a defect but also the type, position, size, etc. of the defect. For example, when the calculation of 2 × S2-S1-S3 is performed, the type of defect is detected from the sign of this calculation value. Thereby, even if a positive defect (tension defect) and a negative defect (chip defect) exist simultaneously in the regular pattern of the inspection object, each can be identified and detected. The position of the defect is determined according to which pattern the image density represents in the second area when the defect is detected. The size of the defect is based on the absolute value of the calculated value when comparing the image densities of the areas including the entire pattern, or when comparing the image densities of the areas obtained by dividing the pattern on the image into a plurality of areas. , It is obtained from the number of areas in which defects are detected for one pattern.
[0019]
According to the defect inspection apparatus according to the present invention, the defect is detected by removing the pseudo defect signal generated by the calculation, so that the false defect is prevented from being misidentified as a true defect, so that the detection accuracy is high. Defect inspection is performed.
[0020]
Among the defect inspection apparatuses according to the present invention, those equipped with a transfer stage perform imaging while continuously moving an inspection object having a regular pattern on the transfer stage. The acquired image information is sequentially sent to the defect detection unit, and the defect detection unit detects a defect based on the image information. Therefore, according to this apparatus, a defect is detected efficiently. Here, when the imaging unit transmits image information at predetermined time intervals, the image information can be acquired at high speed. In addition, when the imaging unit sends image information to the defect detection unit every time the inspection object moves a predetermined distance, even when the moving speed of the inspection object by the transport stage is not uniform in time, the imaging unit is constant. Images can be sampled at distance intervals.
[0021]
In the defect inspection apparatus of the present invention, when the imaging unit acquires one-dimensional image information along the arrangement direction of the regular pattern, the configuration of the imaging unit can be simplified.
[0022]
In the apparatus in which the defect detection unit compares the image density of the region over a plurality of lines, the number of lines including each pattern even when the direction in which the one-dimensional image information is acquired does not match the arrangement direction of the regular pattern By setting so as to compare the image density of the region over the range, it is possible to detect the defect suitably.
[0023]
When the imaging unit includes a line light source and a one-dimensional imaging device, the above-described defect detection is performed by performing imaging by matching the extending direction of the light receiving unit of the one-dimensional imaging device with the arrangement direction of the regular pattern. Is called. At this time, if a means for adjusting the arrangement of the one-dimensional image pickup device is further provided, the image pickup condition can be easily set according to the mode of the regular pattern to be inspected.
[0024]
In addition, when the imaging unit includes a light source that scans the inspection light along the arrangement direction of the regular pattern and a light detection unit, the image information is obtained by sampling the output signal of the light detection unit. The above-described defect detection is performed.
[0025]
Next, when the defect detection unit is provided with a means for performing shading correction, the influence of shading on the image information caused by unevenness in the irradiation of inspection light used for imaging and unevenness in sensitivity of the means for receiving inspection light is reduced. Therefore, erroneous detection is prevented and defect inspection with high detection accuracy is performed.
[0026]
Similarly, when the defect detection unit includes a means for correcting dark current, the influence of the dark current of the imaging unit on the image information is reduced, so that erroneous detection is prevented and defect inspection with high detection accuracy is performed. Is called.
[0027]
【Example】
Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings. In the description of the drawings, the same elements are denoted by the same reference numerals, and redundant description is omitted. Further, the dimensional ratios in the drawings do not necessarily match those described.
[0028]
FIG. 1 is a schematic diagram showing a configuration of a regular pattern defect inspection apparatus according to the present embodiment. This apparatus inspects defects of regular patterns arranged on the inspection object 10 along a predetermined direction with a predetermined repetition period.
[0029]
The plurality of objects to be inspected 10 are mounted on a single-axis stage 12 for conveyance at a certain interval along a direction orthogonal to the arrangement direction of the regular pattern, that is, a conveyance direction indicated by an arrow in the drawing. . The single-axis stage 12 continuously moves the inspection object 10 at a predetermined speed along the conveyance direction of FIG. The uniaxial stage 12 is connected to a control unit 16 via a stage controller 14, and the conveyance speed and the like are adjusted by the stage controller 14. In addition, an encoder 48 is attached to the single-axis stage 12, and the encoder 48 is connected to the control unit 16 so that a data signal indicating the moving distance of the inspection object 10 is output to the control unit 16. It has become.
[0030]
A one-dimensional line light source 18 extending along the arrangement direction of the regular pattern is installed below the uniaxial stage 12 and irradiates the inspection object 10 moving at a predetermined speed with the inspection beam light from below. It is like that. In this embodiment, a transmission light manufactured by Nippon P.I. Co., Ltd. is used as the line light source 18, but a DC-lit fluorescent lamp or the like can also be used.
[0031]
Although not shown in FIG. 1, a diffusion plate that diffuses the irradiation light is installed between the line light source 18 and the uniaxial stage 12 so that the transmitted light that passes through the inspection object 10 is not uneven. It has become. As the diffusion plate, an acrylic plate or a glass plate can be used.
[0032]
A light receiving device 20 including a line sensor camera is installed above the uniaxial stage 12 so as to face the line light source 18 and receives the transmitted light that has passed through the object 10 to be inspected. A regular pattern is imaged. The image data acquired in this way is output as a one-dimensional image signal along the arrangement direction of the regular pattern. As described above, in this embodiment, the line light source 18 and the light receiving device 20 constitute one imaging unit. When acrylic is used for the diffusion plate, infrared light from the line light source 18 is transmitted. Therefore, when the spectral sensitivity of the line sensor element 60a is up to the infrared region, it is better to attach an infrared cut filter. / N can be increased.
[0033]
The light receiving device 20 is connected to a control unit 16 that also has a man-machine interface function, and imaging conditions are set by the control unit 16. In addition, the light receiving device 20 starts imaging a regular pattern in response to a clock signal and a start signal from the control unit 16.
[0034]
A defect detection unit 28 is connected to the light receiving device 20, and the defect detection unit 28 detects a defect of a regular pattern based on image data transferred from the light receiving device 20.
[0035]
The defect detection unit 28 is connected to the control unit 16, and information related to defects in the regular pattern is input from the defect detection unit 28 to the control unit 16. The control unit 16 includes a display device 30a such as a monitor for displaying defect information of the detected regular pattern, and a recording device 30b such as an FD (floppy disk), HD (hard disk), and MO (magneto-optical disk) for recording detection data. External devices 30c such as host computers that receive detection data are respectively connected.
[0036]
FIG. 2 is a schematic diagram showing an aspect of the defect pattern included in the regular pattern on the inspection object 10. If the pattern shown to Fig.2 (a) is made into the normal pattern which becomes a standard, what is shown by FIG.2 (b)-FIG.2 (e) is a pattern which has all a defect. The pattern shown in FIG. 2B is a pattern having a relatively large overall size, and the pattern shown in FIG. 2C is a pattern having a relatively small overall size, and FIG. The pattern shown in FIG. 2 is a pattern having a partial tension, and the pattern shown in FIG. 2E is a pattern having a partial chip.
[0037]
FIG. 3 is a diagram for explaining the configuration of the light receiving means 20 and the control by the control unit 16. Hereinafter, the configuration of the light receiving means 20 will be described in detail with reference to this figure. In FIG. 3, the input of the clock signal and the start signal from the control unit 16 to the light receiving device 20 is omitted.
[0038]
As shown in FIG. 3, the light receiving means 20 includes a shutter 56, a lens 58, a CCD type line sensor camera 60, a rotary stage 61, a height adjustment stage 64, a lens focus controller 62, and a line sensor controller 66. It is configured.
[0039]
The shutter 56 and the lens 58 are connected to the control unit 16 via the lens focus controller 62. The lens focus controller 62 controls the lens focus by driving and controlling the shutter 56 and adjusting the lens position by driving the lens 58.
[0040]
The height adjusting stage 64 is attached to the line sensor camera 60, and adjusts the height of the line sensor camera 60 by its own drive. The rotary stage 61 is also attached to the line sensor camera 60, and corrects the direction of the line sensor camera 60 by its own drive. The rotary stage 61 also functions as a member for attaching the line sensor camera 60 to the height adjustment stage 64. The rotation stage 61 and the height adjustment stage 64 are connected to the control unit 16 via a line sensor / controller 66, and the line sensor / controller 66 controls driving of the rotation stage 61 and the height adjustment stage 64. It is like that.
[0041]
FIG. 4 is a block diagram illustrating configurations of the line sensor camera 60 and the defect detection unit 28. As shown in this figure, a line sensor camera 60 includes a line sensor element 60a that captures a regular pattern on the object to be inspected 10 and outputs an image signal, and an analog filter circuit that performs analog filter processing of the image signal. 60b and an A / D conversion circuit 60c that performs A / D conversion of the image signal subjected to the analog filter processing.
[0042]
The line sensor element 60a converts regular pattern image data obtained by imaging into an image signal representing one-dimensional data along the arrangement direction of the regular pattern, and inputs this to the analog filter circuit 60b. Normally, a process circuit (not shown) having an image signal amplifier circuit or the like is installed between the line sensor element 60a and the analog filter circuit 60b, and the image signal output from the line sensor element 60a is used as the image signal amplifier circuit. The signal is amplified and then input to the analog filter circuit 60b.
[0043]
The analog filter circuit 60b performs an analog filter (low-pass filter or band-pass filter) process on the image signal. This is for removing or suppressing interference fringes generated between the regular pattern and the line sensor element 60a. That is, since the frequency band of the signal corresponding to the regular pattern in the image signal is different from the frequency band of the signal corresponding to the interference fringe, only the frequency band of the interference fringe can be removed by the analog filter processing. Thereby, interference fringes are removed or suppressed.
[0044]
The image signal after the analog filter processing is input to an A / D conversion circuit 60c, where A / D (Analog-to-Digital) conversion processing is performed. This is for preventing the noise induced when the image data is transferred from the line sensor camera 60 to the defect detection unit 28 from affecting the image data. The digitized image data is transferred to the defect detection unit 28 as a digital signal.
[0045]
As shown in FIG. 4, the defect detection unit 28 includes a shading correction circuit 28a that performs shading correction of the line sensor camera 60, a dark current correction circuit 28b that performs dark current correction of the line sensor camera 60, and the like. An adder 28c for performing data addition processing for correction, a spatial filter circuit 28d for performing spatial filter processing such as secondary differentiation, and a LUT (Look Up Table) 28e for performing ternarization at a preset level. And an edge detection circuit 28f for performing edge detection.
[0046]
FIG. 5 is a circuit diagram of the defect detection unit 28 of the present embodiment. The defect detection unit 28 of the present embodiment includes an ALU 70 that multiplies the imaging data and correction data stored in the shading data memory 68 in advance, and data after shading correction and dark current data stored in the dark current data memory 72 in advance. ALU 74 that performs subtraction, ALU 71 that performs product-sum operation on data that has passed through shading correction circuit 28a and dark current correction circuit 28b, 73 that stores the operation result of ALU 71, and product-sum operation on data after dark current correction An ALU 78 to perform, an LUT 28e to perform ternarization at a preset level, and an edge detection circuit 28f that performs edge detection on the output signal of the LUT 28e are provided, and output data of the edge detection circuit 28f is transferred to the control unit 16 It is like that.
[0047]
The spatial filter circuit 28d performs spatial filter processing of several pixels along the scanning direction of the line sensor camera 60. Specifically, the spatial filter circuit 28d performs line processing of several pixels along the scanning direction of the line sensor camera 60 using an ALU 78 (Arithmetic and Logic Unit; computing unit).
[0048]
Next, the operation of the defect inspection apparatus of this embodiment and the defect inspection method using the apparatus of this embodiment will be described.
[0049]
First, prior to starting the inspection, inspection conditions are set. First, the direction of the line sensor camera 60 is corrected. As shown in FIG. 6, the rotation of the rotation stage 61 attached to the line sensor camera 60 is driven, and the line sensor camera 60 is arranged so that the arrangement direction of the regular pattern coincides with the scanning direction of the line sensor camera 60. This corrects the direction of the. Specifically, first, a pseudo pattern identical to the regular pattern to be inspected is created, and a mark hole parallel to the regular pattern and slightly larger than the regular pattern is created on the pseudo pattern. Next, the regular pattern is moved to the position of the mark hole of the pseudo pattern by the uniaxial stage 12 to acquire the image data, and the rotary stage so that the density of the mark hole on the pseudo pattern obtained from this data is all the same. To adjust the position.
[0050]
Next, the height of the line sensor camera 60 is adjusted. Specifically, the height is adjusted so that the repetition period of the regular pattern is an integral multiple of the pixels of the line sensor camera 60. For this purpose, first, a line sensor camera 60 images a jig whose width is known and whose contrast is clear, and acquires data. This data is binarized by the LUT 28e of the defect detection circuit 28, the width is measured, and the actual size value of the length of one pixel is calculated. The LUT 28e is rewritten for setting the inspection condition by the control unit 16, and the setting is different from the case where the data is trinized at the time of the inspection. Thereafter, the control unit 16 sends data to the line sensor controller 66, and drives the height adjustment stage 64 so that the repetition period of the regular pattern is an integral multiple of the length of one pixel, thereby causing the line sensor camera. Adjust the height of 60. Since the height of the line sensor camera 60 at this time is stored by the control unit 16, the next time the same regular pattern is inspected, the line sensor camera 60 automatically moves to that height position. To come. In this embodiment, the regular pattern repetition period is set to be the length of three pixels of the line sensor camera 60.
[0051]
Next, focus control of the line sensor camera 60 is performed. Specifically, the lens focus controller 62 adjusts the position of the lens 58 so that the line sensor camera 60 is defocused by a minute amount more than just focus. The defocus amount is set by the control unit 16 based on the optical waveform data taken into the line sensor camera 60. The lens focus controller 62 adjusts the position of the lens 58 according to the defocus amount.
[0052]
By setting the line sensor camera 60 to a small amount of defocus, interference fringes (moire) generated between the regular pattern and the line sensor camera 60 can be removed or suppressed. The interference fringes appear strongly at the time of just focus and weak at the time of defocus. Therefore, if the line sensor camera 60 is defocused by a minute amount, the interference fringes are removed or suppressed. If interference fringes are included in the image data, the density displacement of the interference fringes may be mistaken as a defect and the inspection accuracy may be reduced. However, if the interference fringes are removed or suppressed as described above, defect inspection is not possible. Accuracy can be increased. Since the focus position of the lens 58 is stored by the control unit 16, the next time the same regular pattern is inspected, the lens 58 automatically moves to that position. When inspecting a regular pattern that does not cause the influence of interference fringes or a regular pattern that can remove the influence of interference fringes almost completely by the analog filter circuit 60b, the line sensor camera 60 is slightly defocused. It is not necessary to adjust to.
[0053]
Next, shading correction data of the line sensor camera 60 is acquired. Specifically, the shading correction circuit 28a and the dark current correction circuit 28b are turned on with the line light source 18 turned on and the shutter 56 opened, and the adder 28c adds the number of times set to stabilize the data. Do the average. The obtained data is sent to the control unit 16. The control unit 16 creates shading correction data from this data and sends it to the shading data memory 68 (FIG. 5) for storage.
[0054]
As will be described later, during the inspection, the shading correction data stored in the shading data memory 68 and the imaging data are multiplied by the ALU 70 (FIG. 5) in the shading correction circuit 28a, and shading correction is performed.
[0055]
Next, dark current correction data of the line sensor camera 60 is acquired. Specifically, the shutter 56 is closed through the lens focus controller 62, the shading correction circuit 28a and the dark current correction circuit 28b are set to through, and the adder 28c performs addition averaging for the number of times set to stabilize the data. . The obtained data is sent to the control unit 16. The control unit 16 sends this data to the dark current data memory 72 (FIG. 5) as dark current correction data for storage.
[0056]
As will be described later, during inspection, subtraction of the dark current correction data stored in the dark current data memory 72 from the imaged data after shading correction is performed by the ALU 74 (FIG. 5) in the dark current correction circuit 28b. Dark current correction is performed.
[0057]
When the above setting work is completed, the inspection is started. First, the conveyance stage 12 is driven, and the line light source 18 is turned on. As a result, the inspection light 10 from the line light source 18 is irradiated onto the inspection object 10 moving on the transport stage 12 from below. The light transmitted through the inspection object 10 is received by the line sensor camera 60 through the lens 58. Thereby, the regular pattern on the inspection object 10 is imaged line by line.
[0058]
The inspection light transmitted through the regular pattern is received by the line sensor element 60a of the line sensor camera 60, and a one-dimensional image signal along the arrangement direction of the regular pattern is output from the line sensor element 60a. This image signal is input to the analog filter circuit 60b where analog filter processing is performed. Thereby, noise that causes a pseudo defect in operation (which will be described later) is reduced, interference fringes are also reduced, and a decrease in S / N can be suppressed. The image signal subjected to the analog filter processing is input to the A / D conversion circuit 60c and A / D converted. As a result, the image signal subjected to the analog filter processing is digitized. The digital image signal thus obtained is transferred to the defect detection unit 28.
[0059]
The transfer timing of the captured image data to the defect detection unit 28 is controlled based on a clock signal and a start signal output from the control unit 16. That is, when the clock signal and the start signal from the control unit 16 are input to the line sensor camera 60, the line sensor camera 60 sends an image signal to the defect detection unit 28.
[0060]
The clock signal is always output from the control unit 16. In addition, the inspection start signal is set in advance on the uniaxial stage 12 as an inspection area before and after the inspection light irradiation area, and a sensor (not shown) attached to the uniaxial stage 12 is set in the inspection area. Is output by detecting that the inspection object 10 has entered. From the output of the inspection start signal to the end of the inspection, a start signal is output every time the inspection object 10 moves a predetermined distance. Specifically, the encoder 48 attached to the single-axis stage 12 measures the movement distance of the inspection object 10 and outputs a signal to the control unit 16 every time the inspection object 10 moves a predetermined distance. The control unit 16 outputs a start signal to the line sensor camera 60 using this signal as a trigger.
[0061]
This method has an advantage that a one-dimensional image can be sampled at a constant distance interval even if the conveyance speed of the single-axis stage 12 is uneven to some extent. However, since the sampling time may not be constant, the amount of charge accumulated on the photocell of the line sensor camera 60 is different at each sampling time, and the image luminance data may vary. In addition, if the line sensor camera 60 is driven with a high-speed clock signal, the output of the encoder 48 may not be counted, so care must be taken.
[0062]
Note that methods other than the above method are also conceivable as methods for determining the transfer timing of image data. In the following, various transfer timing methods are generally described using the concepts of a line sensor system 32 including a line sensor camera 60, a control system 34 including a defect detection unit 28, and a transport system 36 including a stage controller 14. To do.
[0063]
FIGS. 7A to 7C are block diagrams showing the relationship of the line sensor system 32, the control system 34, and the transport system 36 with respect to each method. The first method is a method that does not synchronize with the transport system 36 as shown in FIG. In this method, a start signal is generated in the control system 34 that drives the line sensor system 32, and data is acquired asynchronously with the transport system 36. The inspection start signal is output when a sensor or the like installed in the transport system 36 detects that the inspection object 10 has entered the inspection area in the transport system 36. Subsequent start signals are output at regular time intervals.
[0064]
This method has an advantage that data can be acquired at high speed. Further, since the inspection is performed while continuously moving the object to be inspected 10, a mechanism for stopping the movement is not required, and therefore it is easy to reduce the size of the apparatus and reduce the manufacturing cost. However, if the conveyance speed of the conveyance system 36 is not constant, it is difficult to sample images at a constant distance interval.
[0065]
The second method is a method in which the control system 34 is mainly synchronized as shown in FIG. In this method, an encoder or the like is installed in the transport system 36, the encoder or the like measures the moving distance of the inspection object 10, and outputs a signal to the control system 34 every time the inspection object 10 moves a certain distance. Keep it like that. The control system 34 outputs a start signal to the line sensor system 32 using this signal as a trigger. By repeating the above operation, a start signal is continuously output to the line sensor system 32.
[0066]
This method has an advantage that images can be sampled at constant distance intervals even if there is some unevenness in the conveyance speed of the conveyance system 36. Further, even in this method, since the inspection is performed while continuously moving the object to be inspected 10, there is no need for a mechanism for stopping the movement, and therefore it is easy to reduce the size of the apparatus and reduce the manufacturing cost. Become. However, since the sampling time may not be constant, the amount of charge accumulated on the photocells of the line sensor system 32 differs at each sampling time, and the image luminance data may vary. Further, when the line sensor system 32 is driven by a high-speed clock signal, there is a possibility that the output of the encoder in the transport system 36 cannot be counted.
[0067]
As shown in FIG. 7C, the third method is a method in which the line sensor system 32 mainly performs synchronization. In this system, after the line sensor system 32 acquires one line of data, a movement start signal is output from the control system 34 to the transport system 36, and the transport system 36 performs a certain movement. Further, the control system 34 receives a movement completion signal from the transport system 36 and outputs it to the line sensor system 32 as a start signal. By repeating the above operation, a start signal is continuously output to the line sensor system 32.
[0068]
This method has an advantage that the line sensor system 32 can reliably acquire data because an image is taken in a state where the inspection object 10 is not moving. In addition, there is an advantage that data can be added on the same work. On the other hand, since it is necessary to stop the work for each line, it may not be practical on the inline. In addition, sampling may take time.
[0069]
As described above, each of the above three methods has advantages and disadvantages. Therefore, which method is adopted for the apparatus of the present invention is compared with the advantages and disadvantages according to the conditions of the inspection line and the like. What is necessary is just to determine whether a system is optimal. In this embodiment, a method in which the second and third methods are combined is adopted.
[0070]
Next, processing of the digital image signal transferred to the defect detection unit 28 at the above timing will be described. First, this digital image signal is first input to the shading correction circuit 28a. In the shading correction circuit 28 a, the ALU 70 (FIG. 5) multiplies the image data represented by the digital image signal and the correction data stored in the shading data memory 68. Thereby, the shading correction of the line sensor camera 60 is performed. Thereby, it is possible to remove the influence of the shading on the imaging data due to uneven illumination and uneven sensitivity of the line sensor camera 60.
[0071]
The image signal after the shading correction is input to the dark current correction circuit 28b. In the dark current correction circuit 28b, the ALU 74 (FIG. 5) performs subtraction by subtracting the correction data stored in the dark current data memory 72 from the imaged data after shading correction. Thereby, the dark current component added to the imaging data can be removed, and the dark current correction of the line sensor camera 60 can be performed.
[0072]
The image signal after the dark current correction is input to the spatial filter circuit 28d through the adder 28c. The spatial filter circuit 28d performs a predetermined spatial filter process on the image signal to emphasize the defective portion and make the offset level constant. The image signal output from the spatial filter circuit 28d is input to the LUT 28e and converted into three values, and then input to the edge detection circuit 28f. The edge detection circuit 28 f performs edge detection based on the image signal and transfers the data to the control unit 16.
[0073]
Hereinafter, the data processing will be described in more detail. FIG. 8 is a diagram illustrating a relationship between a regular pattern and pixels. As described above, in this embodiment, the height of the line sensor camera 60 is adjusted so that the repetition cycle of the regular pattern corresponds to 3 pixels. In this figure, one “row” is line data acquired at a time by the line sensor camera 60. As the inspection object 10 is conveyed, line data of each row of A, B, C... Is acquired sequentially. In this figure, a regular pattern to be inspected is schematically shown, but actually, a luminance value or a density value, which is a degree of image brightness, is input to each pixel.
[0074]
The apparatus of this embodiment detects a defect in a regular pattern by comparing density values of one pixel located in the vicinity of each regular pattern and arranged in the same period as the regular pattern repetition period. To do.
[0075]
The comparison operation is performed as a spatial filter process by the spatial filter circuit 28d. The filter pattern of the spatial filter circuit 28d of the present embodiment is a second derivative pattern as shown in FIG. 8 is a combination of symbols A1, A2,... A12... B1, B2,... B12..., C1, C2, combining the density values of each pixel with rows A, B, C. When represented as C12, the spatial filter processing represented by the filter pattern in FIG. 9A is represented by the following arithmetic expression.
[0076]
2 × A4-A1-A7, 2 × A5-A2-A8, 2 × A6-A3-A9, 2 × B4-B1-B7, 2 × B5-B2-B8,… 2 × C6-C3-C9,…
The calculation of 2 × A4-A1-A7 doubles the density value of the pixel located at the upper left of the second regular pattern 102 and is located at the upper left of the first and third regular patterns 101 and 103. This is a value obtained by subtracting the density value of the pixel. If each pixel is expressed as [row, column], each pixel of [A, 1], [A, 4], [A, 7] is arranged with the same period as the regular pattern repetition period. These have a common positional relationship with each pattern. Therefore, the second regular pattern 102 is compared with the first and third regular patterns 101 and 103 positioned before and after the 2 × A4-A1-A7 operation. Other operations are the same as this, and with regard to the density value of one pixel in the vicinity of each pattern, paying attention to three consecutive patterns, among these patterns, the density value for the pattern located in the center, The density values of the patterns before and after that are compared.
[0077]
The above calculation result is 0 when there is no defect in the pattern at the position of the pixel to be compared. On the other hand, as shown in FIG. 10 (a), when there is a tension defect in the pattern at the pixel position [A, 4], the calculation of 2 × A4-A1-A7 is performed as shown in FIG. 10 (b). The value is a positive value. As shown in FIG. 11A, when there is a chip defect in the regular pattern at the pixel position [A, 4], as shown in FIG. 11B, 2 × A4-A1-A7 The calculated value of is a negative value.
[0078]
However, if the three pixels to be compared straddle the boundary between the regular pattern portion (translucency) on the object to be inspected 10 and its peripheral portion (opacity), the nature of the comparison operation An operation value that should be called a pseudo defect is calculated. Specifically, when the case where the pixels [A, 4] and [A, 7] are in the pattern portion and the pixel [A, 1] is in the peripheral portion thereof is considered, Even when there is no defect in the regular pattern 103 of 3 and A4 and A7 are equal, the result A1 in which the pixel [A, 1] is in the non-translucent portion is almost 0, so that 2 × A4-A1-A7 The calculation result is not zero. Normally, such pseudo defects are generated one by one corresponding to the left and right ends of the regular pattern portion.
[0079]
The removal of the pseudo defects due to the opaque portion around the pattern is performed as follows. That is, after the inspection object 10 is imaged before inspection, the imaging data is corrected by the shading correction circuit 28a and the dark current correction circuit 28b, and then binarized by the LUT 28e with the spatial filter circuit 28d being through. Edge detection is performed by the edge detection circuit 28f. Since the spatial filter circuit 28d is made through, the edge detected here corresponds to the image density difference at the boundary between the regular pattern portion (translucent) and the peripheral portion (opaque). This edge position data is sent to the control unit 16 and stored. During the inspection, the control unit 16 determines that only defects appearing between the two edge positions corresponding to the left and right end portions of the regular pattern portion are valid, and performs the above comparison operation based on this. In this way, the above-described pseudo defects can be removed by appropriately setting the measurement range in advance before starting the inspection.
[0080]
Further, in the above-described calculation, calculation-related pseudo defects also occur in patterns before and after the defect pattern. For example, as shown in FIG. 10B, the calculated value of 2 × A7-A4-A10 for the normal pattern located behind the defect pattern is not zero but a negative value. In the case of FIG. 11B, the calculated value of 2 × A7-A4-A10 for the normal pattern located behind the defect pattern is not zero but a positive value.
[0081]
That is, as shown in FIG. 12A, when a pattern has a tension defect, a negative calculation value is generated before and after a positive calculation value indicating the tension defect. These negative calculation values are the result of a comparison operation between a regular pattern having a tension defect and a normal regular pattern around one cycle, and the absolute value is a positive calculation value indicating a tension defect. 1/2. As shown in FIG. 12B, when there is a chip defect, a positive calculation value is generated around one cycle of a negative calculation value indicating the chip defect. These positive calculation values are also generated as a result of a comparison operation between the regular pattern having a chip defect and the normal pattern before and after that, and the absolute value thereof is the absolute value of the negative calculation value indicating the chip defect. 1/2 of this. These pseudo defects are discriminated and removed by the control unit 16.
[0082]
Although there may be cases where tension defects exist continuously, in this case, as shown in FIG. 12 (c), positive calculation values indicating the tension defects are generated at intervals of several cycles. A negative calculation value is generated one cycle before the front positive calculation value and one cycle after the rear positive calculation value. These negative calculation values are also pseudo defects in calculation as in the case of FIG. In this case, the control unit 16 removes the pseudo defect indicating the negative calculation value, and determines that the tension defect exists continuously between the positive calculation values indicating the two tension defects.
[0083]
The same applies to the case where chip defects are continuously present. As shown in FIG. 12D, a minus calculation value indicating a chip defect is generated at a constant interval, and 1 of the minus calculation value at the front is obtained. A positive calculation value is generated before one cycle and after one cycle of the negative calculation value at the rear. These positive calculation values are also pseudo defects in calculation as in the case of FIG. In this case, the control unit 16 removes the pseudo defect indicating a positive calculation value, and determines that the defect exists continuously between the negative calculation values indicating two chip defects.
[0084]
Next, the data after the spatial filter processing is ternarized by the LUT 28e, and then input to the edge detection circuit 28f, where it becomes an object of edge detection. The edge type and edge position data in the data signal detected by the edge detection circuit 28 f are transferred to the control unit 16. Since this data represents the above calculation value, the presence / absence and type of defect (whether it is a tension defect or a chip defect, etc.) is determined from the type of edge, and the defect exists in which pattern from the edge position. Can be determined. Further, when the edges representing the defect are continuous, the defect exists over a continuous series of pixels, and it is determined that the defect is large. In this way, the control unit 16 detects the type, position, and size of the pattern defect based on the data transferred from the edge detection circuit 28f.
[0085]
Further, based on these detection results, the control unit 16 displays a regular pattern defect on the display device 30a, records detection data on the recording device 30b, or detects detection data on the external device 30c as necessary. Or process the data during recording / output.
[0086]
As described above, in the defect inspection apparatus of the present embodiment, by comparing the pixel density located in the vicinity of three regular patterns, not only the presence / absence of a defect but also the type, position, size, etc. of the defect can be determined. It can be detected and a detailed defect inspection is possible. In comparison, the pixel density for the regular pattern to be inspected is doubled, and the pixel density for the preceding and succeeding patterns is subtracted, thereby enhancing the defect and improving the S / N. Thereby, inspection accuracy can be raised.
[0087]
In the apparatus of the present embodiment, a line sensor camera 60 is used as a light receiving device, and a one-dimensional line light source 18 and a one-axis stage 12 are used as a light source, a conveying means, and the like. Is used as a component, it is easy to make the entire apparatus compact and reduce the manufacturing cost.
[0088]
In this embodiment, the CCD line sensor camera 60 is used as the line sensor. However, the present invention is not limited to this. For example, a MOS or PCD line sensor camera may be used. In addition, the control unit 16 according to the present embodiment has a predetermined role based on the first role of controlling the various controllers (62, 66) and the external devices (30a, 30b, 30c) and the output data of the edge detection circuit 28f. However, the apparatus may be configured by preparing a separate control unit for each function.
[0089]
As described above, in this embodiment, the spatial filter processing is performed using the filter pattern shown in FIG. 9A. However, this filter pattern can be rewritten by the control unit 16, thereby providing another space. Filter processing can be performed.
[0090]
Hereinafter, calculation examples representing various spatial filter processes will be shown. The relationship between the regular pattern and the pixel is shown in FIG. Further, as in the case of the embodiment, symbols A1, A2,... A12... B1, B2,. B12..., C1, C2,... C12... And each pixel is represented as [row, column].
[0091]
Example 1
2 × (A4 + A5)-(A1 + A2)-(A7 + A8), 2 × (A5 + A6)-(A2 + A3)-(A8 + A9), 2 × (A6 + A7)-(A3 + A4)-(A9 + A10), 2 × (B4 + B5)-(B1 + B2)-(B7 + B8),… 2 × (C6 + C7)-(C3 + C4)-(C9 + C10) , ...
This compares the sum of density values of two pixels for each pattern. The calculation of 2 × (A4 + A5) − (A1 + A2) − (A7 + A8) doubles the total density of the upper left two pixels of the second regular pattern 102, and then the first and third The same total density of two pixels of the regular patterns 101 and 103 is subtracted. A local region consisting of [A, 1] and [A, 2], a local region consisting of [A, 4] and [A, 5], and a local region consisting of [A, 7] and [A, 8] They are arranged in the same cycle as the regular pattern repetition cycle, and each region has a common positional relationship with each pattern. Therefore, by the calculation of 2 × (A4 + A5) − (A1 + A2) − (A7 + A8), the second regular pattern 102 is converted into the first and third regular patterns 101 and 103 to be compared. Other operations are similar to this. As in the case of the comparison of one pixel performed in the present embodiment, when there is no defect in the pattern in the local region, the calculation result is 0. On the other hand, when the pattern has a defect, the calculation result is 0. Not.
[0092]
Example 2
2 × (A4 + A5 + A6)-(A1 + A2 + A3)-(A7 + A8 + A9), 2 × (B4 + B5 + B6)-(B1 + B2 + B3)-(B7 + B8 + B9 ), 2 × (C4 + C5 + C6)-(C1 + C2 + C3)-(C7 + C8 + C9), ...
This is an example of calculation using the filter pattern of FIG. 9B, and the sum of the density values of three pixels is compared for each pattern. The operation of 2 × (A4 + A5 + A6)-(A1 + A2 + A3)-(A7 + A8 + A9) doubles the total density of the upper left three pixels of the second regular pattern 102, From this, the total density of similar three pixels of the first and third regular patterns 101 and 103 is subtracted. Also in this case, the local regions (the above three pixels) to be compared are arranged with the same cycle as the regular pattern repetition cycle, and each region has a common positional relationship with each pattern. ing. Accordingly, the first and third positions of the second regular pattern 102 positioned before and after the 2 × (A4 + A5 + A6) − (A1 + A2 + A3) − (A7 + A8 + A9) are calculated. Are compared with the regular patterns 101 and 103. The same applies to other operations. If there is no defect in the pattern in the local region, the calculation result is 0. On the other hand, if the pattern has a defect, the calculation result is not 0.
[0093]
Example 3
2 × {(A4 + A5 + A6) + (B4 + B5 + B6) + (C4 + C5 + C6)}-{(A1 + A2 + A3) + (B1 + B2 + B3) + (C1 + C2 + C3)}-{(A7 + A8 + A9) + (B7 + B8 + B9) + (C7 + C8 + C9)}, ...
This is an example of calculation using the filter pattern of FIG. 9C, and the sum of the density values of 9 pixels is compared for each pattern. In the above formula, the total density of 9 pixels in the vicinity of the second regular pattern 102 is doubled, and the same density of 9 pixels of the first and third regular patterns 101 and 103 is subtracted from this. Is. In this case as well, the local regions (9 pixels above) to be compared are arranged with the same cycle as the regular pattern repetition cycle, and each region has a common positional relationship with each pattern. ing. Therefore, the second regular pattern 102 is compared with the first and third regular patterns 101 and 103 located before and after the above operation. If there is no defect in the pattern in the local region, the calculation result is 0. On the other hand, if the pattern has a defect, the calculation result is not 0.
[0094]
When the calculation of Example 3 is performed using data acquired by a one-dimensional line sensor, it is preferable to use a defect detection unit 28 having a circuit configuration as shown in FIG. The spatial filter circuit 28d 'in the defect detection unit 28 has a pipelined circuit configuration. That is, the spatial filter circuit 28d ′ includes a 1H (one horizontal scanning period) delay circuit 76a that performs delay processing of data after dark current correction, and a 1H delay circuit 76b that performs delay processing of data delayed by the 1H delay circuit 76a. The ALU 78 performs a product-sum operation on the data after dark current correction and each data sequentially delayed by the 1H delay circuits 76a and 76b.
[0095]
This spatial filter circuit 28d 'has a buffer of several lines to several tens of lines, and is inspected by using ALU 78 (Arithmetic and Logic Unit) and DDL (Digital Delay Line) 76a and 76b. This is a pipeline processor that performs two-dimensional pipeline processing of three pixels along the moving direction of the object 10 and nine pixels along the scanning direction of the line sensor camera 60.
[0096]
This spatial filter circuit 28d performs two-dimensional spatial filter processing, so that real-time two-dimensional image processing can be performed using the one-dimensional line sensor camera 60. As a result, it is possible to perform movement integration processing for improving S / N and enhancement processing for defect detection on the image data transferred to the defect detection unit 28. Further, even when the inspection object 10 is transported slightly obliquely, the defect can be inspected while reducing the influence thereof, and even when the regular pattern of the inspection object 10 is not uniform, the defect can be detected with high accuracy. Can be inspected. Compared with the case where two-dimensional data is obtained using a two-dimensional camera, the light source, the transport means, and the like are relatively small devices such as the one-dimensional line light source 18 and the one-axis stage 12, so that the device It becomes easy to realize the overall compactness.
[0097]
Furthermore, since the spatial filter process is performed by the pipeline processor and the comparison operation is performed over a plurality of lines (three lines in the case of FIG. 14), the arrangement direction of the regular pattern is a line sensor as shown in FIG. Even when the camera 60 is tilted with respect to the scanning direction, it is possible to detect the defect suitably.
[0098]
When comparing the image density for a plurality of local areas as in this example, the conveyance speed of the uniaxial stage 12 is adjusted prior to the start of inspection, and the regular pattern conveyance direction (movement direction). It is preferable that the length of the local region along the line is set to be an integral multiple of the pixel length.
[0099]
In addition, the spatial filter here refers to a filter configured by a combination of density values of local regions on the original image. For example, a two-dimensional spatial filter with a mask size of 3 × 3 is
[0100]
[Table 1]

[0101]
As shown. In this case, the two-dimensional spatial filter processing means that the density value of each point in the area of the mask size 3 × 3 centering on a certain point (i, j) is weight N
[0102]
[Expression 1]

[0103]
To obtain a new density f (i, j), that is,
[0104]
[Expression 2]

[0105]
This is a process for obtaining.
[0106]
Example 4
2 × (A4 + B4)-(A1 + B1)-(A7 + B7), 2 × (A5 + B5)-(A2 + B2)-(A8 + B8), 2 × (A6 + B6)-(A3 + B3)-(A9 + B9), 2 × (B4 + C4)-(B1 + C1)-(B7 + C7), 2 × (B5 + C5)-(B2 + C2)-(B8 + C8), 2 × (B6 + C6)-(B3 + C3)-(B9 + C9), ...
This is another example of comparing the sum of the density values of two pixels for each pattern. For example, the calculation of 2 × (A4 + B4) − (A1 + B1) − (A7 + B7) doubles the total density of the two pixels on the left side of the second regular pattern 102, and the first and second The total density of the same two pixels of the third regular patterns 101 and 103 is subtracted. In addition, when performing said calculation using the data acquired by the one-dimensional line sensor, it is good to use a pipeline processor as a spatial filter circuit similarly to the case of Example 3.
[0107]
Also in this example, since the comparison operation is performed over a plurality of lines (2 lines), even when the arrangement direction of the regular pattern is inclined with respect to the scanning direction of the line sensor camera 60, the defect is suitably detected. can do.
[0108]
Example 5
2 × (A4 + A5 + B4 + B5)-(A1 + A2 + B1 + B2)-(A7 + A8 + B7 + B8), 2 × (A5 + A6 + B5 + B6)-(A2 + A3 + B2 + B3)-(A8 + A9 + B8 + B9), 2 × (A6 + A7 + B6 + B7)-(A3 + A4 + B3 + B4)-(A9 + A10 + B9 + B10)
This compares the sum of density values of 4 pixels (2 rows and 2 columns) for each pattern. For example, an operation of 2 × (A4 + A5 + B4 + B5) − (A1 + A2 + B1 + B2) − (A7 + A8 + B7 + B8) is performed on four pixels including the second regular pattern 102. The total density is doubled, and the total density of the same four pixels of the first and third regular patterns 101 and 103 is subtracted therefrom. This calculation can also be performed using a pipeline processor as in Example 4.
[0109]
Example 6
2 × (A4 + B4 + C4)-(A1 + B1 + C1)-(A7 + B7 + C7), 2 × (A5 + B5 + C5)-(A2 + B2 + C2)-(A8 + B8 + C8 ), 2 × (A6 + B6 + C6)-(A3 + B3 + C3)-(A9 + B9 + C9), ...
This compares the sum of density values of 3 pixels over 3 lines, that is, 3 pixels along the moving direction of the inspection object 10 for each pattern. For example, the calculation of 2 × (A4 + B4 + C4)-(A1 + B1 + C1)-(A7 + B7 + C7) doubles the total density of the three pixels on the left side of the second regular pattern 102 Then, the total density of similar three pixels of the first and third regular patterns 101 and 103 is subtracted from this. In addition, when performing said calculation using the data acquired by the one-dimensional line sensor, it is good to use the defect detection part 28 of FIG. Moreover, the defect of a pattern like FIG. 14 can be detected suitably also by the calculation of this example.
[0110]
As described above, by comparing the image density in a local region having the same shape on the screen and arranged in a cycle that is an integral multiple of the repetition cycle of the regular pattern, the regular region located in the vicinity of each region is compared. Patterns are compared and regular pattern defects are detected. The width of each area to be compared is preferably within the number of pixels in the arrangement period of the areas (3 pixels in this embodiment) so that the areas do not overlap each other.
[0111]
The defect detection method of this embodiment has the following characteristics according to the number of pixels constituting the local region to be compared and the number of pixels defining the width of the local region to be compared.
[0112]
For example, when the local region is one pixel, there is an advantage that the detection sensitivity of the partial defect is good. However, in this case, it is likely to be affected by noise, and erroneous detection in which the noise is recognized as a defect is relatively likely to occur.
[0113]
When the number of pixels defining the width of the local region is an intermediate number n (1 <n <number of pixels with a repetition period), the density difference of a plurality of pixels is added and compared. It becomes larger and S / N is improved. For this reason, even if it is a partial defect, a misdetection can be suppressed, without reducing the detection sensitivity. Further, even when there is a slight deviation between the arrangement direction of the regular pattern and the scanning direction of the line sensor camera 60, it is possible to detect the defect suitably by comparing with a plurality of lines. However, it is not so easy to determine the number of pixels in the local area.
[0114]
Even when the number of pixels that define the width of the local area is the number of pixels in the regular pattern repetition period (three pixels in this embodiment), as a result of adding the density of a plurality of pixels, the S / N is high and false detection occurs. Few. Further, even when there is a slight deviation between the arrangement direction of the regular pattern and the scanning direction of the line sensor camera 60, it is possible to detect a defect suitably by comparing with a plurality of lines. However, it is somewhat unsuitable for detecting partial defects.
[0115]
In general, n is the number of constituent pixels in the local region to be compared, and the S / N is multiplied by √n by adding the density of these n pixels. Thereby, the influence of noise is reduced and erroneous detection is suppressed.
[0116]
Thus, the feature of defect inspection varies depending on the selection of the local region. Therefore, the width of the local region and the number of pixels constituting the local region may be selected according to the mode of the regular pattern, the mode of the defect, the degree and accuracy of the required defect detection, the time required for detection, and the like. .
[0117]
Next, the arrangement period of the local regions to be compared may be an integer multiple of the number of pixels defining the repetition period of the regular pattern, but the arrangement direction of the regular pattern is the scanning of the line sensor camera 60 as shown in FIG. Considering the case of tilting with respect to the direction, it is preferable that the arrangement period of the local regions is small. This is because the larger the arrangement period, the larger the positional relationship between each local region and each pattern, making it difficult to make an accurate comparison.
[0118]
Example 2
FIG. 15 is a schematic diagram showing the configuration of the defect inspection apparatus of the present embodiment. In this embodiment, an apparatus for imaging a regular pattern to be inspected includes a scanner 80 and a long PMT (Photomultiplier tube) 86.
[0119]
The scanner 80 emits inspection light using a semiconductor laser as a light source, and is disposed above the uniaxial stage 12. The scanner 80 scans the regular pattern at a predetermined speed along the arrangement direction of the regular pattern. The scanner 80 is connected to the defect detection unit 28 via the scanner controller 82, and is driven and controlled by the scanner controller 82.
[0120]
The long PMT 86 has a light receiving surface extending along the arrangement direction of the regular pattern. The long PMT 86 is disposed below the uniaxial stage 12 so as to face the scanner 80, and from the scanner 80 that has passed through the inspection object 10. The inspection light is received.
[0121]
Between the long PMT 86 and the uniaxial stage 12, there is installed a diffusion plate that diffuses the inspection light transmitted through the inspection object 10, so that the light transmitted through the inspection object 10 is not uneven. ing.
[0122]
The long PMT 86 is connected to an A / D conversion circuit 90 that performs A / D conversion on the output signal of the long PMT 86, and the defect detection unit 28 is connected to the A / D conversion circuit 90. ing.
[0123]
FIG. 16 is a block diagram illustrating a configuration of the defect detection unit 28 of the present embodiment. The defect detection unit 28 includes a shading correction circuit 28a that performs shading correction for the long PMT 86, a dark current correction circuit 28b that performs dark current correction for the long PMT 86, and an adder 28c that performs data addition processing for these corrections. And a spatial filter circuit 28d that performs spatial filter processing such as second order differentiation, an LUT 28e that performs ternarization at a preset level, and an edge detection circuit 28f that performs edge detection.
[0124]
The defect detection unit 28 is connected to the control unit 16 that also has a man-machine interface function, and is comprehensively controlled by the control unit 16. Although not shown, the control unit 16 includes a display device such as a monitor that displays defects in a regular pattern, a recording device such as an FD, HD, and MO that records detection data, and an external device such as a host computer that receives the detection data. Each device is connected.
[0125]
Next, the operation of the defect inspection apparatus of this embodiment and the defect inspection method using the apparatus of this embodiment will be described.
[0126]
First, prior to starting the inspection, inspection conditions are set. Specifically, in the same manner as in the first embodiment, shading correction data of the imaging device including the long PMT 86 and the scanner 80 and dark current correction data of the long PMT 86 are acquired. When obtaining shading correction data, the scanner 80 is driven to emit inspection light. When acquiring the dark current correction data, the scanner 80 is stopped so that the inspection light is not emitted.
[0127]
Further, the inspection object 10 is imaged before the inspection, and the digital image signal captured line by line is corrected by the shading correction circuit 28a and the dark current correction circuit 28b, and then passed through the spatial filter circuit 28d. Then, binarization is performed by the LUT 28e, and then edge detection is performed by the edge detection circuit 28f. The edge position data corresponding to the pseudo defect detected by the edge detection circuit 28f is sent to the control unit 16 and stored. Thereby, during the inspection, the control unit 16 validates only the defects appearing between the edge positions corresponding to the pseudo defects, and excludes the range other than the range from the comparison calculation target.
[0128]
When the above setting work is completed, the inspection is started. First, the conveyance stage 12 is driven and the scanner 80 is operated. Thereby, the inspection light from the scanner 80 is irradiated on the inspection object 10 moving on the transport stage 12.
[0129]
As shown in FIG. 17, the light transmitted through the inspection object 10 passes through the diffusion plate and is received by the long PMT 86. While the scanner 80 scans a regular pattern for one line, the long PMT 86 performs sampling to acquire one line of image data. The image data signal is converted into a digital signal by the A / D conversion circuit 90 and then transferred to the defect detection circuit 28.
[0130]
The digital image signal captured line by line is subjected to shading correction by the shading correction circuit 28a, dark current correction by the dark current correction circuit 28b, and then passed through the adder 28c and input to the spatial filter circuit 28d. The The spatial filter circuit 28d performs the spatial filter processing similar to that of the first embodiment, highlights the defective portion, and makes the offset level constant. The image data output from the spatial filter circuit 28d is input to the LUT 28e, ternarized, and then input to the edge detection circuit 28f. The edge detection circuit 28 f performs edge detection based on the image signal and transfers the data to the control unit 16. The controller 16 detects the type, position, and size of the pattern defect based on the data transferred from the edge detection circuit 28f. Based on these detection results, the control unit 16 displays defects in the regular pattern on the display device 30a, records the detection data on the recording device 30b, and outputs the detection data to the external device 30c as necessary. Or process data during recording / output.
[0131]
The defect inspection apparatus of the present embodiment can detect not only the presence / absence of a defect but also the type, position, size, etc. of the defect by comparing the pixel densities located in the vicinity of the three regular patterns. And detailed defect inspection is possible. In comparison, the pixel density for the regular pattern to be inspected is doubled, and the pixel density for the preceding and succeeding patterns is subtracted, thereby enhancing the defect and improving the S / N. Thereby, inspection accuracy can be raised.
[0132]
Furthermore, since the defect inspection apparatus of this embodiment receives inspection light by a point sensor called the long PMT 86, there is an advantage that no interference fringes (moire) are generated compared to the case of using a line sensor, and a high-precision defect. Inspection is easier. In addition, the setting of inspection conditions is easier than in the case where a line sensor is used, and therefore efficient defect inspection is possible.
[0133]
【The invention's effect】
According to the defect inspection apparatus of the present invention, the image densities S1, S2 and S3 of the predetermined three regions are compared, and thereby the first and second regions located in the vicinity of these regions in the regular pattern on the image are compared. Since the second and third patterns are compared, it is possible to easily detect not only the presence / absence of a defect but also the type, position, size, etc. of the defect.
[0134]
According to the defect inspection apparatus of the present invention that detects a defect by removing a pseudo defect signal generated by an operation, it is possible to prevent the false defect from being misidentified as a true defect, and thus has high detection accuracy. Defect inspection can be performed.
[0135]
According to the defect inspection apparatus of the present invention, which includes a transfer stage, image information is sequentially transferred to the defect detection unit by performing imaging while continuously moving an inspection object having a regular pattern by the transfer stage. It is sent out and a defect can be detected efficiently.
[0136]
In the defect inspection apparatus of the present invention, if the imaging unit acquires one-dimensional image information along the arrangement direction of the regular pattern, the configuration of the imaging unit can be simplified, Miniaturization and reduction of manufacturing costs can be achieved.
[0137]
When the defect detection unit compares the image density of a region over a plurality of lines, the number of lines including each pattern even when the direction in which the one-dimensional image information is acquired does not match the arrangement direction of the regular pattern By setting so as to compare the image density of the region over the range, it is possible to detect the defect suitably.
[0138]
If the defect detection unit is equipped with a means for correcting shading, the influence of shading caused by unevenness in the irradiation of inspection light used for imaging and unevenness in the sensitivity of the means for receiving inspection light is reduced, resulting in false detection. Therefore, it is possible to perform defect inspection with high detection accuracy.
[0139]
Similarly, if the defect detection unit is provided with means for correcting dark current, the influence of the dark current of the imaging unit on the image information is reduced and erroneous detection can be prevented, so that defect detection with high detection accuracy is possible. It can be performed.
[Brief description of the drawings]
FIG. 1 is a schematic diagram illustrating a configuration of a defect inspection apparatus for a regular pattern according to a first embodiment.
FIG. 2 is a schematic diagram showing an embodiment of a regular pattern on the inspection object 10;
FIG. 3 is a diagram for explaining a configuration of a light receiving unit 20 and control by a control unit 16;
4 is a block diagram showing configurations of a line sensor camera 60 and a defect detection unit 28. FIG.
FIG. 5 is a circuit diagram of the defect detection unit 28 according to the first embodiment.
6 is a diagram showing a relationship between an arrangement direction of a regular pattern and a scanning direction of the line sensor camera 60. FIG.
FIG. 7 is a block diagram showing the relationship among a line sensor system 32, a control system 34, and a transport system 36 for various image data transfer timing systems.
FIG. 8 is a diagram illustrating a relationship between a regular pattern and pixels. It is a schematic diagram for demonstrating the relationship between the regular pattern of a to-be-inspected object and the pixel of a line sensor camera in case the regular pattern of to-be-inspected object inclines.
FIG. 9 is a diagram showing a filter pattern of the spatial filter circuit 28d.
FIG. 10 is a diagram for explaining a calculation result in the defect detection unit when the regular pattern has a tension defect.
FIG. 11 is a diagram for explaining a calculation result in the defect detector 28 when a regular pattern has a chip defect.
FIG. 12 is a diagram schematically showing a calculation result in the defect detection unit 28 when various defects are present in the regular pattern.
FIG. 13 is a circuit diagram showing a pipeline type spatial filter circuit 28d ′.
14 is a diagram showing a state in which the arrangement direction of the regular pattern is inclined with respect to the scanning direction of the line sensor camera 60. FIG.
FIG. 15 is a schematic diagram illustrating a configuration of a defect inspection apparatus according to a second embodiment.
FIG. 16 is a block diagram illustrating a configuration of a defect detection unit according to the second embodiment.
17 is a schematic diagram showing a relationship between a scanner 80 and a long PMT 86. FIG.
[Explanation of symbols]
DESCRIPTION OF SYMBOLS 10 ... Test object, 12 ... Single axis stage, 14 ... Stage controller, 16 ... Control part, 18 ... Line light source, 20 ... Light receiving means, 28 ... Defect detection part, 28a ... Shading correction circuit, 28b ... Dark current correction Circuit 28c ... Adder 28d ... Spatial filter circuit 28e ... LUT 28f ... Edge detection circuit 30a ... Display device 30b ... Recording device 30c ... External device 32 ... Line sensor system 34 ... Control system 36 ... transport system, 44 ... stage controller, 48 ... encoder, 56 ... shutter, 58 ... lens, 60 ... line sensor camera, 60a ... line sensor element, 60b ... analog filter circuit, 60c ... A / D conversion circuit, 61 ... Rotation stage 62 ... Lens focus controller 64 ... Height adjustment stage 66 ... Line sensor controller Roller, 68 ... Shading data memory, 70 ... ALU, 71 ... ALU, 72 ... Dark current data memory, 73 ... Memory, 74 ... ALU, 76a and 76b ... 1H delay circuit, 78 ... ALU, 80 ... Scanner, 82 ... Scanner Controller, 86 ... long PMT, 90 ... A / D conversion circuit.

Claims (12)

An apparatus for inspecting defects of a plurality of regular patterns arranged at a predetermined repetition period along a predetermined direction on an inspection object,
An imaging unit that images the regular pattern and obtains image information of the regular pattern;
Based on the image information, the image densities S1, S2, and S3 of three regions arranged along the arrangement direction of the regular pattern at an integer multiple of the repetition period of the regular pattern on the image are obtained. A defect detection unit for detecting defects in the regular pattern by comparing the image densities S1, S2 and S3;
Equipped with a,
The defect detection unit calculates 2 × S2-S1-S3 using the image densities S1, S2, and S3, and detects defects in the regular pattern based on the calculation result . Defect inspection equipment. The regular pattern defect inspection apparatus according to claim 1 , wherein the defect detection unit detects a defect by removing a pseudo defect signal generated by the calculation.   The regular pattern defect inspection apparatus according to claim 1, further comprising a transfer stage that moves the inspection object along a predetermined direction. The transfer stage is for continuously moving the inspection object,
4. The regular pattern defect inspection apparatus according to claim 3 , wherein the imaging unit sends the image information to the defect detection unit at a predetermined time interval. The transfer stage is for continuously moving the inspection object,
The regular image defect inspection apparatus according to claim 3 , wherein the imaging unit sends the image information to the defect detection unit every time the inspection object moves a predetermined distance.   The regular pattern defect inspection apparatus according to claim 1, wherein the imaging unit acquires one-dimensional image information along a predetermined direction. The defect detection unit simultaneously processes the one-dimensional image information of a plurality of lines along a direction orthogonal to the arrangement direction of the regular pattern, thereby obtaining an image density of the three regions over the plurality of lines. The defect inspection apparatus for a regular pattern according to claim 6, wherein: The imaging unit is a line light source having a light emitting part extending in the arrangement direction of the regular pattern and irradiating the inspection object with inspection light emitted from the light emitting part, and a light receiving part extending in a predetermined direction. The regular pattern defect inspection apparatus according to claim 6, further comprising: a one-dimensional imaging device having a light-receiving element that receives the inspection light that has passed through the inspection object. 9. The regular pattern defect inspection apparatus according to claim 8 , further comprising means for adjusting an arrangement of the one-dimensional image pickup device. The imaging unit irradiates the inspection object with inspection light and scans the inspection light along the arrangement direction of the regular pattern; and a light receiving surface extending along the arrangement direction of the regular pattern The regular pattern defect inspection apparatus according to claim 6, further comprising: a light detection unit including: a light detection unit that receives the inspection light that has passed through the inspection object.   The regular pattern according to claim 1, wherein the defect detection unit includes means for performing shading correction on the image information, and detects defects based on the image information corrected thereby. Defect inspection equipment.   2. The regular pattern according to claim 1, wherein the defect detection unit includes means for performing dark current correction on the image information, and detects a defect based on the image information corrected thereby. Defect inspection equipment.

Images