565681 五、發明說明（1 ) [技術領域] 本發明係關於測量測定對象面之凹凸形狀的表面形狀測 €方法及其裝置，尤其和使用白色光或單色光以非接觸方 式測量測定對象表面之技術相關。 [習知技術] 以往’此種裝置，利用以雷射等單色光之干涉來測量薄 膜或光學元件等精密加工品之凹凸形狀的方法之表面形狀 測定裝置廣爲人知。傳統之表面形狀測定方法，利用光束 分光鏡將來自單色光源之單色光分成照射至測定對象面之 單色光’及照射至對照面的單色光，並利用自兩面分別反 射回來之各單色光的干涉現象，測量測定對象面之凹凸形 狀。 亦即，使光束分光鏡上下移動，可對應對照面至光束分 光鏡之距離，及光束分光鏡至測定對量面之距離的差使其 產生千涉現象，然後測量產生該干涉現象之單色光（以下 簡稱爲『干涉光』）的強度。 此時，假設分別從測定對象面及對照面反射回來之兩單 色光的相位差爲4、從對照面之高度爲h、N爲整數時，h 可以下述（1)式來表示。 h = λ /2 X {(0 /2π) + N} ...(1) 又，干涉光之相位變動量的單位爲π/2，以使其相位差 爲0、π/2、π、3π/2之4次變化方式上下移動光束分光鏡 。單色光之波長爲λ時，代表光束分光鏡會在0、λ /8、 565681 五、發明說明（2) λ/4、3λ/8上移動。若此時之干涉光強度爲a、B、C、D 時，則相位差4可以下述（2)式表示。 0 = tan·1 {(A-C)/(B-D)} ...(2) 利用前述（2)式求取之相位差0中，因具有2πΝ之不定 性，故在實際求取h時，需要該（1)式中之右側的ν項。 因此，在測定對象面之高度十分平滑且從水平方向觀察時 之商度爲連繪變化之假設下’實施相位連續（un-lapping)處 理，在特定N之後，即可求取從對照面之高度h。相位連 續法有例如單純之近接連續法、以及較爲複雜之演算法的 MST(Minimum Spanning Tree:最小生成樹）法等。又，依據 此相位差求取高度之方法被稱爲『相位偏移干涉法（Phase Shift Interferometry)』（以下簡稱爲『PSI法』）。 然而，此種PSI法係假設高度爲連續變化，故以波長而 言，測定對象必須具有比該薄膜或光學元件等之測定對象 面的凹凸更爲平滑之條件。 因此，PSI法只適合用於測定對象面爲平滑面時，對於 其他例如半導體凸塊或屬加工面等表面段差較大之測定對 象面，則無法進行測量。 本發明有鑑於此，故本發明之目的就是提供對表面段差 較大之測定對象面亦可進行精度良好之測量的表面形狀測 定方法及其裝置。 [發明之槪述] 565681 五、發明說明（3) 本發明者爲了達成該目的，經過精密硏究而獲得下述之 真知識。 表面形狀測定裝置除了使用前述PSI法之裝置以外，尙 有使用『垂直掃描干涉法（Vertical Scanning Interferometry)』（以下稱稱爲『VSI法』）之裝置爲眾所皆 知。利用此方法，將來自白色光源之白色光分成照射於測 定對象面之白色光、及照射於對照面之白色光，並測量兩 面分別反射回來之各白色光所形成之干涉光的強度。PSI 法時，以使其相位差爲0、π/2、π、3π/2之4次變化方式 上下移動光束分光鏡，V SI法則是以特定間隔上下移動光 束分光鏡，實施特定間隔之干涉光的抽樣。然後，以求取 干涉光之強度値變化的波形（以下稱爲『干涉圖』）爲最大 之位置，來求取測定對象面之高度。 實際上，特定間隔之抽樣資料十分離散，很難以良好精 度求取干涉光強度爲最大之位置。因此，依據干涉圖之包 絡曲線、或本發明者先前提出之『日本國特開2001-066122號公報』發明之特性涵數爲最大之位置，來測量測定 對象面之高度。又，本說明書中之『干涉光』，不只是干涉 圖而已，尙包括該包絡曲線及特性涵數等在內。 利用該VSI法時，和PSI法不同，因未假定高度爲連續 變化，故可實施該半導體凸塊或金屬加工面等表面段差較 大之測定對象面的測量。 然而，VSI法時，如前面所述，不易求取干涉光強度爲 565681 五、發明說明（4 ) 胃大時之位置，且有必須花費相當多時間來實施求取包絡 曲'線或特性涵數相關之干涉光強度的演算之問題。 另一方面，和依據單色光之干涉光強度來求取高度之 PSI法不同，VSI法係依據白色光之千涉光強度來求取高 度。因爲PSI法爲了使干涉光之相位在0、π/2、π、3π/2 移動，波長λ必須爲已知，故PSI法會採用如雷射之單色 光’而VSI法若使用單色光，則干涉性會較大，且求取最 大之干涉光的強度變化（尖峰）會較爲平坦，故測定精度愈 變差，故相反的，有較大尖峰値之白色光最適合VSI法。 本發明者發現，以採用不同光源之PSI法及VSI法的組 合，可以達成本發明之目的。以此發現爲基礎之本發明， 係具有下述構成。 亦即，本發明相關之表面形狀測定方法，係依據光干涉 來測量測定對象面之凹凸形狀的表面形狀測定方法，其特 徵爲具有：第1步驟，以特定間隔變動對照面及該測定對象 面之相對距離；第2步驟，每次在該第1步驟變動該相對距 離時，測量對該對照面及該測定對象面照射光所得之干涉 光的強度；第3步驟，在該第2步驟中測得之各該干涉光強 度當中，將干涉光強度爲最大之該相對距離推算爲推算距 離；第4步驟，依據接近該第3步驟推算之該推算距離的複 數個相對距離時之各干涉光強度，求取干涉光相位；以及 第5步驟，依據該第3步驟推算之該推算距離、及該第4 步驟求取之該干涉光相位，求取該測定對象面之高度。 565681 五、 發明說明（5) 依 據 本 發明之表面形狀測定方法，在第1步驟中以特定 間 隔 變 動 對照面及測定對象面之相對距離，在第2步驟 中 則在 第 1 步驟每次變動相對距離時，測量對照面及測定 對 象 面 之干 涉光強度。在第3步驟中，則將該第2步驟測 得 之 各 干 涉 光強度當中之最大干涉光強度的相對距離推算 爲 推 算 距 離 。從第1步驟到第3步驟，係執行從干涉光之 強 度 變 化 測 量測定對象面高度之垂直掃描干涉法（VSI法）的 步 驟 〇 第 4 步 驟則依據接近第3步驟推算之推算距離的複數 個 相 對 距 離 時之各干涉光強度，求取干涉光之相位，在第 5 步 驟 中 則依據第3步驟推算之該推算距離，及第4步 驟 求取 之 干 涉光相位，求取測定對象面之高度。此第4步 驟 及 第 5 步 •驟，係執行依據各干涉光強度求取干涉光相位 且 依 據 該 相位測量測定對象面高度之相位偏移干涉法（psi法） 的 步 驟 〇 因 此 y 依據第3步驟推算之推算距離可確定測定對象 面 之 局 度 而沒有必要實施以求取測定對象面之高度爲目 的 之 相位 連 續（un-lapping)處理，且不但可以測量測定對象 面 爲 平 滑 面 者，亦可測量具有較大表面段差之表面者。又 > 在 第 3 步 驟求取推算距離後，又在第5步驟求取測定對 象 面 之 筒 度 ，故可實施精度良好之測量。結果，可以良好 精 度 測 量 表 面段差較大之測定對象面。 在 第 4 步驟中，爲了依據接近該第3步驟推算之前述 -7- 推 565681 五、發明說明（6 ) 算距離的複數個相對距離時之各干涉光強度來求取干涉光 相位上，只要和傳統之PSI法同樣具有3個以上之相對距 離的各干涉光強度相關資料即可，其代表方法有3點資料 法、4點資料法、及5點資料法等。其中，4點資料法有 下述具體方法。亦即’依據靠近推算距離之4個相對距離 的各干涉光強度，求取0、π/2、π、3π/2時之干涉光的相 位變動量，同時，將該時之4個干涉光強度分別當做A、 B、C、D，且以（/)表示必須求取之相位時，則可從0 = tanKA-C^B-D)}之公式來求取相位0。 在第5步驟中，依據第3步驟推算之推算距離，及第4 步驟求取之干涉光相位，求取測定對象面之高度時，有下 述具體方法。亦即，N爲整數，λ爲在第2步驟照射至對 照面及測定對象面之光的波長，h爲依據第3步驟推算之 推算距離的推算高度，（^爲第4步驟求取之相位，h2爲第 5步驟必須求取之測定對象面時，在第5步驟中，選擇整 數N使利用1ι2=λ /2 X {(0 /2π) + N}求取之測定對象面高 度h2最接近推算高度，然後，再以選取之Ν來求取測 定對象面之高度h2。 又，可使用第2步驟測得之干涉光強度在第4步驟中求 取干涉光相位，亦可利用第2步驟測得之干涉光強度實施 波形還原，然後，再依據該波形還原結果，在第4步驟求 取干涉光相位。 前者時，在第1步驟中改變相對距離之特定間隔在第2 565681 五、 發明說明（7) 步 驟 中 照 射於對照面及測定對象面之光的波長爲 λ 時 爲 λ /8 5 或 在 N爲整數時則爲（N/2 + 1/8) X λ。例 如 干 涉 光 之 相 位 變動量爲π/2而在〇、π/2、π、3π/2上 移 動 時 可 以 直 接 應用對應相位變動量之λ /8。因此，無 需 再 復 原 波 形 Ο 後 者 時 ，在第1步驟中改變相對距離之特定間 隔 在 第 2 步 驟 中 照射於對照面及測定對象面之光的波長 爲 λ 時 會 大於 λ /8 且小於奈奎斯（Nyquest)間隔時，可以依抽 樣 定 理 使用 第 2 步驟測得之干涉光強度實施波形還原， 再 依 據 該 波 形 還 原 之結果，並在第4步驟求取干涉光相位 ， 而在 第 1 步 驟 中 改變相對距離之特定間隔若大於奈奎斯 間 隔 時 可 以 依 帶 通抽樣定理使用第2步驟測得之干涉光 強 度 實 施 波 形 還 原 ，再依據該波形還原之結果，並在第4 步 驟 求取 干渉光 相 位。此時，因爲特定間隔大於λ /8，故和 特 定 間 隔 爲 λ /8 時相比，前者只要較少之干涉光強度的抽 樣 個 數 ’ 而 可 縮 短干涉光強度之抽樣時間。 又 y 亦可在不同於第1步驟之第4步驟中，會 再 度 變 動 對 照 面 及 測定對象面之相對距離，並依據該變動 後 之 各相 對 距 離 以及接近第3步驟推算之推算距離的複 數 個 相 對 距 離 之 各干涉光強度，求取干涉光相位。再度變 動 相 對 距 離 5 對 在 後述第2步驟中之光照射的光源，以及在: 第 4 步 驟 中 光 照 射的光源不同時十分有效。 又 ， 在 第2步驟中照射至對照面及測定對象面 -9 - 之光 5 最 565681 五、發明說明（8) 好爲例如白色光。白色光時，求取最大時之千涉光強度變 化（尖峰）會較爲明顯，在第2步驟後之第3步驟中可以較 容易求取干涉光強度爲最大之相對距離。 又，第2步驟係VSI法之步驟之一，第4步驟以後，則 爲PSI法之步驟之一，故最好不直接採用第2步驟之照射 用白色光，且在第4步驟照射頻帶較第2步驟照射用白色 光之頻帶更狹窄的白色光或單色光，並依據頻帶較第2步 驟照射用白色光之頻帶更狹窄之白色光或單色光相關的各 干涉光強度，求取干涉光相位。因爲在第4步驟中求取相 位時，波長必須爲已知，且此種光爲頻帶較第2步驟照射 用白色光之頻帶更爲狹窄之白色光或單色光。 又，亦可使用同一光源來照射第2步驟之白色光、及第 4步驟之白色光或單色光，但將第2步驟中照射之白色光 限制於狹窄頻帶，然後將其應用於第4步驟之照射上，此 外，亦可使用不同光源來照射第2步驟之白色光、及第4 步驟之白色光或單色光。 又，本發明相關之表面形狀測定裝置具有：光源，產生 照射至測定對象面及對照面之光；變動機構，以特定間隔 變動該測定對象面及對照面之相對距離；攝影機構，產生 之干涉條紋會隨該光照射之測定對象面及對照面間的相對 距離變動而變化，同時會對該測定對象面進行攝影；抽樣 機構，以取得該攝取之測定對象面上複數個特定部位之干 涉光強度値爲目的，以該特定間隔依序讀取對應依該變動 -10- 565681 五、發明說明（9) 機構之該測定對象面及對照面之相對距離變動而變化的特 定部位干涉光強度値；以及演算機構，依據以該由抽樣機 構取得之各特定部位的複數個強度値一各千涉光強度値群 ，求取該複數個特定部位之各別高度，並利用該各別高度 來測量該測定對象面之凹凸形狀；之表面形狀測定裝置， 且其特徵爲：該演算機構具有將以該抽樣機構取得之干涉 光強度値群當中之最大干涉光強度的該相對距離推算爲推 算距離之第1處理，依據接近該推算距離之複數個相對距 離之各干涉光強度求取干涉光相位之第2處理，以及依據 推算所得之該推算距離及求取之該干涉光相位來求取該特 定部位之高度的第3處理。 依據本發明之表面形狀測定裝置，光源會產生光，光則 會分別照射至測定對象面及對照面。變動機構會使光照射 之測定對象面及對照面之相對距離產生變動。攝影機構則 會攝取對應分別由測定對象面及對照面反射之光的光路差 而變化之干涉條紋，同時實施測定對象面及對照面之攝影 ，故可掌握對應測定對象面之凹凸形狀而產生之干涉條紋 及干涉條紋之變化。抽樣機構以取得攝取之測定對象面上 複數特定部位之干涉光強度値爲目的，以前述特定間隔依 序讀取對應依該變動機構之該測定對象面及對照面之相對 距離變動而變化的特定部位千涉光強度値。演算機構會依 據抽樣機構取得之各特定部位的複數個強度値一各干涉光 強度値群，分別求取複數個特定部位之各別高度，測量測 -11- 565681 五、發明說明（1〇 ) 定對象面之凹凸形狀。 具體而言，演算機構會先執行第1處理，將以抽樣機構 取得之干涉光強度値群當中之最大干涉光強度的相對距離 推算爲推算距離，然後執行第2處理，依據接近推算距離 之複數個相對距離之各干涉光強度求取干涉光相位，再執 行第3處理，依據推算所得之推算距離，及求得之干涉光 相位來求取特定部位之高度。 亦即，第1處理係相當於本發明之表面形狀測定方法的 第3步驟，第2處理係相當於本發明之表面形狀測定方法 的第4步驟’第3處理則係相當於本發明之表面形狀測定 方法的第5步驟。又，變動機構最好採用本發明之表面形 狀測定方法的第1步驟，而變動機構及光源則最好採用本 發明之表面形狀測定方法的第2步驟。 如上面所述，利用本發明之表面形狀測定裝置可以使本 發明之表面形狀測定方法獲得良好執行。 [圖式之簡單說明] 第1圖係本發明第1實施例之表面形狀測定裝置的槪要 構成方塊圖。 第2圖係說明干涉條紋之機構的說明圖。 第3圖係說明以VSI法求取推算高度之說明圖。 第4圖係說明以VSI法求取推算高度之說明圖。 第5圖係第1實施例之表面形狀測定方法的處理流程圖 -12- 565681 五、發明說明（11) 第6圖係第2實施例之表面形狀測定裝置的槪要構成方 塊圖。 [發明之最佳實施形態] 下面爲以解決傳統問題點爲目的之形態。 <第1實施例> 參照圖面說明本發明第1實施例。第1圖係本發明第1 實施例之表面形狀測定裝置的槪要構成方塊圖，第2圖係 說明干涉條紋之機構的說明圖，第3圖及第4圖係說明以 VSI法求取推算高度之說明圖，第5圖則係第1實施例之 表面形狀測定方法的處理流程圖。 如第2圖所示，從半透明鏡16至對照面15之距離爲L1 ，而和半透明鏡I 6相距L1距離之位置上的平面則爲平面 E。又，以試料台50爲基準，從該處至平面E之高度h爲 干涉計之位置，而試料之測定對象面31上之點P的高度 爲hP 〇 第1實施例之表面形狀測定裝置的構成如第1圖所示， 具有對半導體晶片、玻璃基板、或金屬基板等之測定對象 物30上形成之微細圖案照射特定頻帶之白色光的光學系 元件1、及控制光學系元件1之控制系元件2。 光學系元件1之構成上具有：白色光源1 〇，產生照射於 測定對象面3 1及對照面1 5之白色光；準直透鏡1 1，使來 自白色光源1 0之白色光成爲平行光；帶通濾波器〗2，限制 頻帶較來自白色光源10之白色光更狹窄之白色光（在第i -13- 565681 五、發明說明（12) 實施例中爲中心波長600nm、帶寬20nm);半透明鏡13, 將通過帶通濾波器1 2之白色光、或未通過帶通濾波器1 2 而直接從白色光源1 〇照射之白色光朝測定對象物30之方 向反射，並可使來自測定對象物30之方向的白色光通過； 物鏡14，對在半透明鏡13反射之白色光進行集光；半透明 鏡1 6，將通過物鏡1 4之白色光分成反射至對照面1 5之對 照光、及通過測定對象面31之測定光，同時，再度集中 在對照面1 5反射回來之參照光、及在測定對象面31反射 回來之測定光，並產生干涉條紋；作動器1 7，可執行物鏡 1 4、對照面1 5、及半透明鏡1 6之上下左右驅動；成像透鏡 1 8，使由對照光及測定光匯整而成之白色光形成圖像；以 及CCD攝影機1 9，攝取干涉條紋及測定對象面3 1之影像 〇 · 白色光源1 〇可以爲例如白色光燈，產生頻帶較寬之白 色光。此白色光源10所產生之白色光會因準直透鏡11而 變成平行光，並入射至帶通濾波器1 2。 帶通濾波器1 2係只讓特定頻帶之白色光通過的濾波器 ，裝設於白色光源10至CCD攝影機19之光路上。然而， 其設置位置最好係位於白色光源1 〇至將該白色光源1 〇之 白色光分成對照面1 5之對照光及測定對象面3 1之測定光 之位置間的光路上。本實施例中，係將其裝設於準直透鏡 1 1及半透明鏡1 3間之光路上。帶通濾波器1 2方面，利用 如中心波長600nm、頻寬20nm之帶通型光學干涉濾波器 -14- 565681 五、發明說明（13) 等。入射至此帶通濾波器12之較寬頻帶的白色光，只有 頻帶較狹窄之特定頻帶的白色光會通過帶通濾波器12。帶 通濾波器1 2相當於本發明之頻帶限制機構。 又，本實施例中，帶通濾波器1 2可利用下述之濾波器 切換部24進行切換，在後述之步驟S 1〜S5中，可利用濾 波器切換部24使帶通濾波器12退至準直透鏡11及半透 明鏡1 3間之光路之外，白色光源1 0產生之白色光可以不 通過帶通濾波器12而維持較寬之頻帶。在下述之步驟S6 以後，可利用濾波器切換部24使帶通濾波器1 2位於準直 透鏡1 1及半透明鏡1 3間之光路上，而白色光源1 0產生 之白色光通過帶通濾波器1 2而限制爲較狹窄頻帶之白色 光。 又，在本說明書中，將未通過帶通濾波器12而由白色 光源10直接照射之白色光定義爲『第1白色光』，同時 ，將通過帶通濾波器12之較狹窄頻帶之白色光定義爲『 第2白色光』。 半透明鏡1 3可使第1白色光或第2白色光朝測定對象 物30之方向反射，且會讓從測定對象物30之方向回來的 白色光通過。在此半透明鏡13反射之第1/第2白色光會 入射至物鏡1 4。 物鏡1 4係將入射之白色光朝焦點p方向集光之透鏡。 利用此物鏡1 4集光之第1 /第2白色光會通過對照面1 5而 到達半透明鏡1 6。 -15- 565681 五、發明說明（14) 半透明鏡16會將物鏡14集光之第1/第2白色光，分成 以在對照面1 5反射爲目的而在半透明鏡1 6反射之對照光 、以及以在測定對象面31反射爲目的而通過半透明鏡1 6 之測定光，同時，會以再度匯整該對照光及測定光來產生 干涉條紋。到達半透明鏡1 6之白色光，會分成在半透明 鏡1 6反射之對照光、及通過半透明鏡1 6之測定光，該對 照光會到達對照面1 5，該測定光則會到達測定對象面3】 〇 在對照面1 5反射之對照光會到達半透明鏡1 6，然後， 該對照光會再度在半透明鏡1 6反射。 通過半透明鏡1 6之測定光，會朝焦點P集光，而在測 定對象面31反射。此反射之測定光會到達半透明鏡1 6並 通過該半透明鏡1 6。 半透明鏡1 6會再度匯集對照光及測定光。此時，光路 會因爲對照面1 5及半透明鏡1 6之間的距離L1、以及半透 明鏡1 6及測定對象面3 1之間的距離L2不同而產生差異 。對照光及測定光會對應此光路差而產生互相干涉，並產 生干涉條紋。在產生此千涉條紋之狀態下的白色光，會通 過半透明鏡1 3，並由成像透鏡1 8形成圖像，然後入射至 CCD攝影機19。 作動器1 7係以變化光學系元件1內之對照面1 5及半透 明鏡16間之固定距離L1、以及半透明鏡16及測定對象面 3 1之可變距離L2的距離差爲目的，而使物鏡14、對照面 -16- 565681 五、發明說明（15) 15、及半透明鏡16可在垂直相交之3軸方向移動的裝置 ，其構成上，係依據CPU20之指示而在X、γ、Z軸上驅 動物鏡1 4、對照面1 5、及半透明鏡1 6。在本實施例中， 係使物鏡1 4、對照面1 5、及半透明鏡1 6分別移動，然而 ，亦可使光學系元件1整體移動，例如，使載置著測定對 象物30之工作台（圖上未標示）在垂直3方向上移動。作動 器1 7相當於本發明之變動機構。 又，本發明之變動機構亦可以如3軸驅動型伺服馬達等 驅動機構來實施上下左右之驅動。 CCD攝影機1 9會攝取產生干涉條紋之狀態下的白色光 ，以及測定光映出而位於測定對象面31之焦點P附近之 圖像。控制系元件2會收集該攝取之圖像資料。又，利用 作動器1 7使物鏡14、對照面1 5、及半透明鏡1 6在上下 左右移動。利用前述上下方向之移動可改變距離L1及距 離L2的差距。利用此方式，干涉條紋可對應距離L1及距 離L2之距離差而變化。在各後述抽樣間隔（第i實施例中 爲λ /8)，利用CCD攝影機19攝取干涉條紋之變化及測定 對象面31之圖像，並利用控制系元件2收集該圖像資料 。CCD攝影機1 9相當於本發明之攝影機構。又，抽樣間 隔相當於本發明之特定間隔。 控制系元件2會綜合控制整體表面形狀測定裝置，其構 成上，則係具有：CPU20，執行特定之演算處理；記億體21 ，儲存以CPU20逐步收集之圖像資料、及CPU20之演算 1 L--- 565681 五、發明說明（16) 結果等各種資料；輸入部22’輸入抽樣間隔或其他設定資 訊之滑鼠及鍵盤等；監視器23 ’顯示測定對象面3丨之圖像 等；濾波器切換部24，依據cpU2〇之指示切換帶通濾波器 12;之電腦系統。又，CPU20相當於本發明之抽樣機構及 演算機構。 CPU20亦即所謂之中央處理裝置，除了控制CCD攝影 機1 9、記憶體21、及光學系元件1之作動器1 7以外，尙 會依據含有CCD攝影機1 9攝取之干涉條紋的測定對象面 31圖像資料，執行測量測定對象面31之凹凸形狀的處理 。後面將詳細說明此處理。又，CPU20連結著監視器23、 及鍵盤或滑鼠等之輸入部22，操作者可在觀察顯示於監視 器23之操作畫面的同時，從輸入部22執行各種設定資訊 的輸入。又，結束測定對象面31之測量後，監視器23會 以數値或圖像來顯示測定對象面3 1之凹凸形狀。 在後述之步驟S1〜S5中，濾波器切換部24爲使白色光 源10產生之白色光可不通過帶通濾波器12而直接入射至 測定對象物30，亦即，使第1白色光直接照射於測定對象 物30上，而使帶通濾波器1 2移至準直透鏡π及半透明 鏡1 3間之光路之外，而在下述之步驟S 6以後，爲了使只 有頻帶較狹窄之白色光源1〇產生的白色光可通過帶通濾 波器1 2並入射至測定對象物30，而使帶通濾波器〗2位於 準直透鏡Π及半透明鏡1 3間之光路。亦即，濾波器切換 部24係使帶通濾波器1 2可在準直透鏡1 1及半透明鏡i 3 -18- 565681 五、發明說明（17) 間之光路內外移動的裝置。和CCD攝影機1 9相同，濾波 器切換部24之構成上，係依據CPU20之指示而在X、Y 、Z軸上驅動帶通濾波器1 2。 以下，參照第5圖之流程圖，說明本實施例之表面形狀 測定裝置整體的處理。 步驟S1(設定爲第1白色光） 首先，在濾波器切換部24處於使帶通濾波器12位於準 直透鏡1 1及半透明鏡1 3間之光路之外的狀態下，白色光 源10產生白色光。因此，白色光源1〇產生之白色光會未 通過帶通濾波器12而維持較寬之頻帶，並入射至測定對 象物30及對照面15。因爲帶通濾波器12位於準直透鏡 1 1及半透明鏡1 3間之光路之外，而實施第1白色光之設 定。從設定後至步驟S6之第2白色光設定爲止，會對測 定對象面3 1及對照面1 5照射第1白色光。 步驟S2(以特定間隔變動相對距離） CPU20會對作動器17提供開始移動指示，使原本位於 特定測定位置之光學系元件1開始朝Z軸方向移動。作動 器1 7會使物鏡14、對照面1 5、及半透明鏡1 6在Z軸方 向上只移動預設之距離。因此，對照面1 5及測定對象面 3 1之相對距離會變動。此步驟S2相當於本發明之第1步 驟。 在本實施例中，係依波長λ之1 / 8 ( λ / 8)的特定間隔一亦 即抽樣間隔來變動相對距離之方式使其在Ζ軸方向上移動 -19- 565681 五、發明說明（18) 。又’最初之抽樣位置爲h〇、抽樣間隔爲△時，則會在z 軸 向之各抽樣間隔△的、+△、hG + 2 △、…、1!。+ (11-1)八 上移動，並針對含CCD攝影機1 9攝取之干涉條紋在內之 測定對象面31的圖像資料，依各抽樣間隔附與編號。如 第3圖所示，抽樣位置時，附與i。之編號，抽樣位置 h，△時則附與之編號。其後亦依此附與編號。 又，若分別從對照面1 5及測定對象面3 1之兩平面反射 之干涉光強度（干涉圖）爲g(h)，則各抽樣間隔爲g(hQ)、 g(hQ+A )、g(hQ + 2A )、…、g(hQ + nA )，而獲得干涉光強度 値群。 步驟S3(是否可取得特定之資料數？） 若無法從干涉光強度群獲得特定之資料數，則回到步驟 S2，在步驟S2依特定間隔（抽樣間隔）變動相對距離，並在 各抽樣間隔對對照面1 5及測定對象面3 1照射白色光，取 得干涉光強度。若可取得特定之資料數，則進入步驟S4。 此步驟S3相當於本發明之第2步驟。 步驟S4(求取包絡曲線） 干涉光強度値群中，干涉光強度變化最大之抽樣位置爲 測定對象面之高度，然而，實際上各特定間隔之抽樣資料 會離散，故不易從g(h)求取干涉光強度爲最大之位置。因 此，依據干涉圖之包絡曲線、或本發明者先前提出之『日 本國特開2001 -0661 22號公報』發明之特性涵數爲最大之 位置，來測量測定對象面之高度。 -20· 565681 五、發明說明（19 ) 在本實施例中，係求取包絡曲線並將該包絡曲線爲最大 時之位置當做推算高度，此推算高度爲h。 從求取包絡曲線之各干涉光強度値群減去干涉光強度値 群之平均値，再將實施減算之資料進行平方。所以，利用 帶通濾波器1 2之平滑化而可得到第4圖所示之包絡曲線。 步驟S5(求取推算高度） 獲得包絡曲線後，求取此包絡曲線爲最大之位置的推算 高度h。此時之圖像編號若爲i,，則推算高度h,變成 b + J△。此步驟S5相當於本發明之第3步驟，亦相當於本 發明之第1處理。 步驟S6(設定爲第2白色光） 濾波器切換部24使帶通濾波器1 2位於準直透鏡1 1及 半透明鏡13間之光路。白色光源10產生之白色光會通過 帶通濾波器1 2而使受到較狹窄之頻帶限制的白色光入射 至測定對象物30及對照面1 5。因爲帶通濾波器1 2位於準 直透鏡11及半透明鏡13間之光路上’而實施第2白色光 之設定。從設定後至步驟SU之顯示爲止，會對測定對象 面3 1及對照面1 5照射第2白色光。 步驟S7(接近推算高度之抽樣位置的析出） 析出接近推算高度之4個抽樣位置。本實施例中’係析 出 h,、1^ + Δ、Μ + 2Δ、h, + 3△，然而，亦可以爲 、hi、hi+Δ、h, + 2△、或 1^-2^、1ι「Δ、hi、h，△、或 h广 3 Δ ' h,-2A 、1^·Δ 、。 -21 - 565681 五、發明說明（2〇) 又’抽樣間隔△爲λ / 8，干涉光相位之變動量則對應於 π/2。然後，設定於析出之位置並進入步驟S8。 步驟S8(求取干涉光相位） h、hi+Δ、1^ + 2Δ、h + 3△時之各干涉光強度爲A、B 、C、D。又，必須求取之干涉光的相位差爲4，則相位差 0可利用A、B、C、D以該（2)式求取。此步驟S8相當於 本發明之第4步驟，亦相當於本發明之第2處理。 步驟S9(求取測定對象面之高度） 最後求取之測定對象面的高度爲h2。若取相位4後，以 該（1)式之左邊的h爲h2，選擇使高度h2最接近推算高度 h之整數N，再從選取之N求取測定對象面之高度h2。此 步驟S9相當於本發明之第5步驟，亦相當於本發明之第3 處理。 步驟S10(全部特定位置都結束？） 至全部特定位置都結束爲止，CPU20會重複執行步驟 S 1〜S9之處理，求取全部特定位置之高度。 又，實際上會以電視攝影機（2次元CCD檢測器）進行檢 測，故步驟S 1 0只會重複計算步驟。 步驟S11(顯示） CPU20會將特定位置之高度資訊顯示於監視器23上， 顯示依據各特定位置之高度資訊的3次元或2次元圖像。 操作者可以觀察這些顯示，掌握測定對象物30之測定對 象面3 1的凹凸形狀。 -22- 565681 五、發明說明（21) 依據第1實施例，在步驟S2會以特定間隔之抽樣間隔 △改變對照面及前述測定對象面之相對距離，在步驟S3 中，每次以抽樣間隔△變動時，會分別測量白色光在對照 面及測定對象面上之干涉光的強度。在步驟S 5中，會求 取步驟3中測得之各干涉光強度値群當中干涉光強度爲最 大之推算高度。在本實施例中，會在步驟S4求取包絡曲 線，而在步驟S5中求取該包絡曲線爲最大之推算高度。 步驟S 1〜步驟S5之至測量干涉光強度爲止，係垂直掃描 干涉法（VSI法）之步驟。 在步驟S8中，依據步驟S5推算之推算高度求取干涉光 之相位差4，在步驟S 9中，依據步驟S 5推算之推算高度 、及步驟S8求取之干涉光相位差4求取測定對象面之高 度。步驟S9中，依據各干涉光強度求取干涉光相位，而 至依據該位相測量測定對象面之高度爲止，係相位偏移干 涉法（PSI法）之步驟。 因此，可以依據步驟S5推算之推算高度來確定測定對 象面之高度，無需實施以求取測定對象面之高度爲目的之 位相連續（un-lapping)處理，且不但可以測量測定對象面爲 平滑面者，亦可測量具有較大表面段差之表面者。又，因 爲在步驟S5求取推算高度後，又在步驟S9求取測定對象 面之高度，故可實施精度良好之測量。結果，表面段差較 大之測定對象面亦可實施精度良好之測量。 抽樣間隔△爲λ /8時’ 4個抽樣位置之各干涉光相位變 -23- 565681 五、發明說明（22) 動爲對應λ /8之π/2，在步驟S8中，依據各抽樣間隔△之 各干涉光強度求取干涉光相位。 此時，要以π/2之干涉光相位變動量在〇、π/2、π、3π/2 移動，因可直接利用對應相位變動量之λ /8，故無需重新 實施波形還原，而可利用已求取之各抽樣間隔λ /8的各干 涉光強度來求取干涉光相位。 又，照射於對照面及測定對象面之光係白色光，和單色 光相比，白色光之干涉性較低，因求取最大之干涉光強度 之變化（尖鋒）較明顯，故在步驟S5中，較易求取干涉光強 度爲最大之推算高度。 又，步驟S3係所謂VSI法之步驟的一部份，而步驟S 8 以後則爲所謂PSI法之步驟的一部份，故不直接利用步驟 S1〜S6照射之白色光（第1白色光），步驟S7〜S10係照射 頻帶較步驟S1〜S6照射之白色光狹窄的白色光（第2白色 光），並依據第2白色光相關之各干涉光強度來求取干涉 光相位。要在步驟S8求取相位，波長λ必須爲已知，因 爲此種光係頻帶較步驟S3照射之第1白色光更狹窄之白 色光。又，在本實施例中，第1/第2白色光因非單色光， 故無法完全決定波長λ，然而，因第2白色光之頻帶較爲 狹窄，故可將頻帶內之特定波長視爲已知之波長λ，且決 定λ /8之抽樣間隔△亦可。 <第2實施例> 其次，說明本發明第2實施例。和第1實施例相同之部 -24- 565681 五、發明說明（23) 份附與相同符號’並省略其說明。第2實施例之表面形狀 測定裝置，如第6圖所示’其光學系元件1具有不同於白 色光源1 〇之其他光源的白色光源10b ’且’控制系元件2 具有切換白色光源10,10b之光源切換部25，取代第1實 施例之控制系元件2內的濾波器切換部24。在構成上，由 此白色光源l〇b產生之白色光的頻帶會較白色光源10產 生之白色光的頻帶更爲狹窄。亦即，白色光源1〇產生之 白色光爲第1白色光，白色光源l〇b產生之白色光則爲第 2白色光。因爲表面形狀測定方法之處理係和第1實施例 相同，故省略其說明。 和第1實施例不同之處，爲沒有以將白色光限制於較狹 窄頻帶爲目的之帶通濾波器12,然而，最好具有帶通濾波 器12 ° 本發明並不限於前述實施形態，亦可爲下面所示之變形 實施形態。 (1) 前述第1、第2實施例中，光學系元件1係所謂『 邁克生干涉儀』，然而，亦可使用被稱爲『（？P22L21-22)MIRAN干涉儀』之光學系元件（參照『日本國特開 200 1 -066 1 22 號公報』）。 (2) 前述第1、第2實施例中，係以析出hl、hjA、 1^ + 2Δ、h! + 3△之4點的所謂『4點資料法』來求取相位 ’然而，只要有3個以上之干涉光強度的相關資料即可， 其代表如3點資料法、5點資料法。 -25 - 565681 五、發明說明（24) (3 )則述第1、第2實施例中，V S I法係使用頻帶較寬 之白色光（第1白色光）’而PSI法則使用狹窄頻帶之白色 光（第2白色光），然而，若波長；^能獲得某種程度之限定 ’則PSI法亦可使用和VSI法相同的白色光。 又’例中之PSI法爲狹窄頻帶之白色光，然而，亦可爲 單色光。單色光時，因爲波長λ會完全已知，故更容易決 定抽樣間隔△。又，第1實施例時，帶通濾、波器1 2係利 用只會使單色光通過之濾波器，而第2實施例時，則在白 色光源1 〇以外，尙具有可產生單色光之光源。 (4)目丨j述弟1貫施例中’抽樣間隔爲；（/ 8，然而，亦可 以更大。 未實施波形還原，而N爲整數時，抽樣間隔亦可以爲 (Ν/2+1/8)Χ又。因Ν λ /2係對應1周1，故只要使此抽樣 間隔對應相位，即可實施各不同周期之抽樣。亦即，Ν爲 1 時，只要對應 〇、5 λ /8(= λ /2+ λ /8)、5 久 /4(= λ + λ /4) 、…，貝!]干涉光位相會在 〇、 5π/2( = 2π + π/2) 、 5π( = 4π + π)、 …移動。當然，Ν亦可以爲1以外，各抽樣之Ν亦爲隨機 變動。此時，干涉光位相亦會以π/2之間隔在0、π/2、π、 3π/2變動，故可直接利用對應位相變動量之λ /8，而無需 重新實施波形還原。 還原波形時，可以有例如下述之方法。利用抽樣定理時 ，抽樣間隔只有大於λ /8、小於奈奎斯間隔即可，利用帶 通抽樣定理時，抽樣間隔只要大於奈奎斯間隔即可。無論 -26- 565681 五、發明說明（25) 何種情形，利用抽樣定理或帶通定理雖然都需要實施波形 還原，但因抽樣間隔大於λ /8，和抽樣間隔爲λ /8時相比 ’只需取得較少之干涉光強度的抽樣數，而可縮短干涉光 強度之抽樣時間。 (5) 前述第1實施例中，以包絡曲線當做干涉光強度， 然而，亦可以採用干涉圖，此外，亦可將本發明者先前提 出之『日本國特開2001 -0661 22號公報』發明的特性涵數 當做干涉光強度，然後求取特性涵數爲最大之位置即可。 (6) 前述第1、第2實施例中，VSI法及PSI法使用不同 的光。因此，最好在步驟S2中變動對照面及測定對象面 之相對距離（L1+L2)並獲得干涉光強度，同時在步驟S6中 再度變動對照面及測定對象面之相對距離並獲得干涉光強 度。當然，在PSI法中使用和VSI法相同之光時，則無需 實施再度變動來獲得干涉光強度。 [產業上之利用可能性] 如前面所述，本發明之表面形狀測定方法及其裝置，係 利用光干涉來測量測定對象物之凹凸形狀，且其不但適用 於半導體晶片或液晶顯示器用玻璃基板等精密加工品，亦 適用於半導體凸塊或金屬加工面等表面段差較大之測定對 象面。 [元件符號之說明] 1 光學系元件 2 控制系元件 -27- 565681 五、發明說明（26) 10、10 b 白色光源 11 準直透鏡 12 帶通濾波器 13 > 16 半透明鏡 14 物鏡 15 對照面 17 作動器 18 成像透鏡 19 CCD攝影機 20 CPU 21 記憶體 22 輸入部 23 監視器 24 濾波器切換部 25 光源切換部 30 測定對象物 31 測定對象面 50 試料台 -28 -565681 V. Description of the invention (1) [Technical Field] The present invention relates to a method and a device for measuring a surface shape of a concave-convex shape of a measurement target surface, and in particular, to measure the surface of a measurement target in a non-contact manner using white light or monochromatic light Technology-related. [Conventional Technology] Conventionally, such a device has been widely known as a surface shape measuring device for measuring the uneven shape of a precision processed product such as a thin film or an optical element by the interference of monochromatic light such as laser. The traditional method for measuring the surface shape uses a beam splitter to separate the monochromatic light from a monochromatic light source into a monochromatic light that is irradiated to the measurement target surface and a monochromatic light that is irradiated to the control surface, and is used to reflect the light from the two surfaces. The interference phenomenon of monochromatic light measures the uneven shape of the measurement target surface. That is, moving the beam splitter up and down can correspond to the distance from the control surface to the beam splitter, and the distance between the beam splitter and the measuring surface to make it involved, and then measure the monochromatic light that produces the interference phenomenon. (Hereinafter referred to as "interfering light"). In this case, assuming that the phase difference between the two monochromatic light reflected from the measurement target surface and the reference surface is 4, and the height from the reference surface is h and N is an integer, h can be expressed by the following formula (1). h = λ / 2 X {(0 / 2π) + N} ... (1) In addition, the unit of the phase variation of the interference light is π / 2 so that the phase difference is 0, π / 2, π, Move the beam splitter up and down by 4 times of 3π / 2. When the wavelength of monochromatic light is λ, it means that the beam splitter will move at 0, λ / 8, 565681. 5. Description of the invention (2) λ / 4, 3λ / 8. When the interference light intensity at this time is a, B, C, or D, the phase difference 4 can be expressed by the following formula (2). 0 = tan · 1 {(AC) / (BD)} ... (2) The phase difference 0 obtained by using the above formula (2) has an uncertainty of 2πN. Therefore, when actually calculating h, you need The ν term on the right side of the formula (1). Therefore, under the assumption that the height of the measurement object surface is very smooth and the quotient when viewed from the horizontal direction is a continuous drawing change, 'un-lapping' is performed. After specifying N, the value from the control surface can be obtained. Height h. The phase continuity method includes, for example, a simple close-continuous method, and a more complex algorithm, the Minimum Spanning Tree (MST) method. The method of obtaining the height based on this phase difference is called "Phase Shift Interferometry" (hereinafter referred to as "PSI method"). However, this PSI method assumes that the height changes continuously. Therefore, in terms of wavelength, the measurement target must have conditions smoother than the unevenness of the measurement target surface of the film or optical element. Therefore, the PSI method is only suitable when the measurement target surface is a smooth surface. For other measurement target surfaces with large surface segment differences, such as semiconductor bumps or processing surfaces, the measurement cannot be performed. In view of this, the present invention aims to provide a method and a device for measuring a surface shape that can also perform a highly accurate measurement on a measurement target surface having a large surface segment difference. [Description of the invention] 565681 V. Description of the invention (3) In order to achieve the purpose, the inventors obtained the following true knowledge through precise research. In addition to the device using the aforementioned PSI method, a surface shape measuring device is known that uses a "Vertical Scanning Interferometry" (hereinafter referred to as "VSI method"). With this method, white light from a white light source is divided into white light that is irradiated on the measurement target surface and white light that is irradiated on the control surface, and the intensity of interference light formed by each white light reflected from both sides is measured. In the PSI method, the beam splitter is moved up and down so that the phase difference is 4 times of 0, π / 2, π, and 3π / 2. The V SI rule is to move the beam splitter up and down at a specific interval to perform interference at a specific interval. Sampling of light. Then, the height of the measurement target surface is obtained by obtaining the position where the waveform of the intensity 値 change of the interference light (hereinafter referred to as the "interference pattern") is the maximum. In fact, the sampling data at specific intervals is very discrete, and it is difficult to find the position where the interference light intensity is the largest with good accuracy. Therefore, the height of the surface to be measured is measured based on the envelope curve of the interference diagram or the position where the characteristic eigenvalue of the invention disclosed in "Japanese Patent Application Laid-Open No. 2001-066122" previously proposed by the inventor is the largest. The "interfering light" in this specification is not only an interference pattern, but also includes the envelope curve and characteristic quotient. When the VSI method is used, unlike the PSI method, since the height is not assumed to change continuously, it is possible to perform measurement on a measurement target surface having a large surface difference such as the semiconductor bump or the metal-machined surface. However, in the VSI method, as described earlier, it is not easy to obtain the interference light intensity of 565681. V. Description of the invention (4) The position when the stomach is large, and it must take a considerable amount of time to implement the envelope curve or characteristic. Number-related calculation of interference light intensity. On the other hand, unlike the PSI method, which determines the height based on the intensity of interference light of monochromatic light, the VSI method uses the intensity of white light to determine the height. In order for the PSI method to move the phase of the interference light between 0, π / 2, π, and 3π / 2, the wavelength λ must be known. Therefore, the PSI method uses monochromatic light such as lasers, and the VSI method uses mono Light, the interference will be greater, and the maximum change in the intensity of the interference light (peak) will be flat, so the measurement accuracy will become worse, so on the contrary, white light with large peaks is most suitable for the VSI method . The inventors have found that the combination of the PSI method and the VSI method using different light sources can achieve the purpose of the invention. The present invention based on this finding has the following constitution. That is, the surface shape measurement method according to the present invention is a surface shape measurement method for measuring the uneven shape of a measurement target surface based on optical interference, and is characterized in that it has a first step of changing the control surface and the measurement target surface at specific intervals. Relative distance; in the second step, each time the relative distance is changed in the first step, the intensity of interference light obtained by irradiating the control surface and the measurement target surface with light is measured; in the third step, in the second step Among the measured interfering light intensities, the relative distance at which the interfering light intensity is maximum is calculated as an estimated distance; in the fourth step, the interfering lights at a plurality of relative distances near the estimated distance calculated in the third step are calculated. Intensity, the phase of the interference light is obtained; and in the fifth step, the interference light phase calculated in the third step and the interference light phase obtained in the fourth step are used to obtain the height of the measurement target surface. 565681 V. Description of the invention (5) According to the surface shape measuring method of the present invention, in the first step, the relative distance between the control surface and the measurement target surface is changed at a specific interval, and in the second step, the relative distance is changed each time in the first step. At the distance, the interference light intensity of the control surface and the measurement target surface is measured. In the third step, the relative distance of the maximum interference light intensity among the interference light intensities measured in the second step is calculated as the estimated distance. From the first step to the third step, the steps of the vertical scanning interference method (VSI method) for measuring the height of the target surface from the change in the intensity of the interference light are performed. The fourth step is based on the plurality of estimated distances approximated to the third step. The interference light intensity at the relative distance is used to obtain the phase of the interference light. In the fifth step, the estimated distance calculated in the third step and the interference light phase obtained in the fourth step are used to obtain the height of the measurement object surface. . This 4th step and 5th step are the steps of phase shift interference method (psi method) for obtaining the phase of the interference light based on the intensity of each interference light and measuring the height of the object surface based on the phase measurement. Therefore, y is based on the 3rd step The estimated distance of the step estimation can determine the locality of the measurement target surface, and it is not necessary to implement an un-lapping process for the purpose of obtaining the height of the measurement target surface, and not only can the measurement target surface be a smooth surface, but also Can measure the surface with larger surface segment difference. And > After obtaining the estimated distance in the third step, and obtaining the degree of measurement of the object's barrel in the fifth step, it is possible to carry out a measurement with good accuracy. As a result, it is possible to measure the measurement target surface having a large surface difference with good accuracy. In the fourth step, in order to obtain the interference light phase based on the above-mentioned -7- 565565 which is approximated in the third step, and the description of the invention (6) the interference light intensity when calculating a plurality of relative distances of the distance, as long as Similar to the conventional PSI method, it is sufficient to have three or more relative light intensity related data of relative distances. The representative methods include the 3-point data method, the 4-point data method, and the 5-point data method. Among them, the four-point data method has the following specific methods. That is, based on the intensity of the interference light at the four relative distances close to the estimated distance, the phase variation of the interference light at 0, π / 2, π, and 3π / 2 is obtained, and the 4 interference lights at that time are calculated. When the intensity is regarded as A, B, C, D, and the phase must be obtained by (/), then the phase 0 can be obtained from the formula of 0 = tanKA-C ^ BD)}. In the fifth step, there are the following specific methods to determine the height of the measurement target surface based on the estimated distance estimated in the third step and the interference light phase obtained in the fourth step. That is, N is an integer, λ is the wavelength of the light irradiated to the reference surface and the measurement target surface in the second step, h is the estimated height of the estimated distance calculated according to the third step, and (^ is the phase obtained in the fourth step) When h2 is the measurement target surface that must be obtained in the fifth step, in the fifth step, the integer N is selected so that the height h2 of the measurement target surface obtained by using 1ι2 = λ / 2 X {(0 / 2π) + N} is the largest. Approach the estimated height, and then use the selected N to obtain the height h2 of the measurement target surface. The interference light intensity measured in the second step can be used to obtain the interference light phase in the fourth step. The interference light intensity measured in the step is subjected to waveform reduction, and then, according to the waveform reduction result, the interference light phase is obtained in step 4. In the former case, the specific interval for changing the relative distance in step 1 is on the 2nd 565681. Description of the invention In the step (7), the wavelength of the light irradiated on the reference surface and the measurement target surface is λ / 8 5 when the wavelength is λ or (N / 2 + 1/8) X λ when N is an integer. For example, interference light The phase variation is π / 2 and it is at 0, π / 2, π, 3π / 2 Λ / 8 corresponding to the amount of phase variation can be directly applied during motion. Therefore, there is no need to restore the waveform. 0 In the latter case, the specific distance of the relative distance is changed in the first step. The light irradiated on the control surface and the measurement target surface in the second step When the wavelength of λ is larger than λ / 8 and smaller than the Nyquest interval, you can use the interference light intensity measured in the second step to perform waveform restoration according to the sampling theorem, and then use the waveform restoration result according to the result of the waveform restoration. Step 4 is to obtain the phase of the interference light, and if the specific interval for changing the relative distance in step 1 is greater than the Nyquist interval, you can use the bandpass sampling theorem to implement the waveform reduction using the interference light intensity measured in step 2 and then according to the The result of waveform reduction, and the phase of the dry light is obtained in step 4. At this time, because the specific interval is larger than λ / 8, compared with the case where the specific interval is λ / 8, the former requires fewer samples of interference light intensity. Number 'to shorten the sampling time of the interference light intensity. In the same step as in the fourth step of the first step, the relative distances between the control surface and the measurement target surface are changed again, and the interferences of the plurality of relative distances close to the estimated distance estimated in the third step are changed according to the relative distances after the change and the interference distance. The light intensity is used to determine the phase of the interference light. Varying the relative distance 5 again is very effective for the light source irradiated with light in the second step described later, and the light source irradiated with light in the fourth step is not effective at the same time. In addition, in the second step, the light irradiated to the reference surface and the measurement target surface -9-5 is the most 565681 5. The invention description (8) is, for example, white light. In the case of white light, the maximum light intensity change (spike) at the time of obtaining the maximum will be obvious. In the third step after the second step, it is easier to obtain the relative distance at which the interference light intensity is the maximum. In addition, the second step is one of the steps of the VSI method. After the fourth step, it is one of the steps of the PSI method. Therefore, it is better not to directly use the white light for the second step of irradiation, and to irradiate the frequency band in the fourth step. The second step irradiates white light or monochromatic light with a narrower frequency band of white light, and obtains the interference light intensity related to the white light or monochromatic light with a narrower frequency band than that of white light in step 2 Interfering light phase. This is because when the phase is obtained in the fourth step, the wavelength must be known, and this light is white light or monochromatic light with a narrower frequency band than that of the white light irradiated in the second step. It is also possible to use the same light source to illuminate the white light in step 2 and the white light or monochromatic light in step 4, but limit the white light irradiated in step 2 to a narrow frequency band, and then apply it to the fourth In the step of irradiation, in addition, different light sources may be used to illuminate the white light in step 2 and the white light or monochromatic light in step 4. The surface shape measuring device according to the present invention includes: a light source that generates light irradiating the measurement target surface and the reference surface; a change mechanism that changes the relative distance between the measurement target surface and the reference surface at a specific interval; and a photography mechanism that generates interference The fringe will change as the relative distance between the measurement target surface and the control surface irradiated by the light changes, and the measurement target surface will be photographed at the same time; the sampling mechanism will obtain the interfering light at a plurality of specific locations on the measurement target surface taken in For the purpose of intensity 目的, sequentially read the interference light intensity at a specific location corresponding to the change in accordance with the change -10- 565681 V. Description of the invention (9) Changes in the relative distance between the measurement target surface and the reference surface of the mechanism 値; And a calculation mechanism, based on the plurality of intensities of each specific location obtained by the sampling mechanism (one thousand light intensity groups), obtain the respective heights of the plurality of specific locations, and use the respective heights to measure An uneven shape of the measurement target surface; and a surface shape measuring device, characterized in that the calculation mechanism has The relative distance of the maximum interference light intensity among the groups of interference light intensity obtained from the structure is estimated as the first processing of the estimated distance, and the second of the interference light phase is obtained based on the interference light intensity of a plurality of relative distances close to the estimated distance. And a third process of obtaining the height of the specific part based on the estimated distance obtained from the estimation and the phase of the interference light obtained. According to the surface shape measuring device of the present invention, the light source generates light, and the light is irradiated to the measurement target surface and the control surface, respectively. The changing mechanism changes the relative distance between the measurement target surface and the reference surface irradiated with light. The photographing mechanism will pick up interference fringes corresponding to the light path difference of the light reflected by the measurement target surface and the reference surface, and simultaneously perform the photography of the measurement target surface and the control surface, so it can grasp the uneven shape corresponding to the measurement target surface. Interference fringes and changes in interference fringes. The sampling mechanism aims to obtain the interference light intensity 値 of a plurality of specific parts on the measurement target surface that is taken, and sequentially reads the specific corresponding to the change of the relative distance change of the measurement target surface and the control surface at the specific interval at the aforementioned specific interval. Part of the light intensity involved. The calculation mechanism will obtain the respective heights of the plurality of specific locations based on the multiple intensities of each specific location obtained by the sampling agency, and the intensity of each interfering light, respectively. The uneven shape of the target surface. Specifically, the calculation mechanism first performs the first processing, and estimates the relative distance of the maximum interference light intensity among the interference light intensity groups obtained by the sampling mechanism as the estimated distance, and then executes the second processing based on the complex number close to the estimated distance. The interference light phase of each relative distance is used to obtain the interference light phase, and then the third process is performed, and the height of the specific part is obtained based on the estimated distance obtained from the calculation and the obtained interference light phase. That is, the first treatment corresponds to the third step of the surface shape measurement method of the present invention, and the second treatment corresponds to the fourth step of the surface shape measurement method of the present invention. The third treatment corresponds to the surface of the present invention. Step 5 of the shape measurement method. It is preferable that the changing mechanism uses the first step of the surface shape measuring method of the present invention, and the changing mechanism and the light source preferably use the second step of the surface shape measuring method of the present invention. As described above, the surface shape measuring method of the present invention can be performed well using the surface shape measuring apparatus of the present invention. [Brief description of the drawings] Fig. 1 is a block diagram of a schematic configuration of a surface shape measuring device according to the first embodiment of the present invention. Fig. 2 is an explanatory diagram illustrating a mechanism of interference fringes. FIG. 3 is an explanatory diagram for obtaining the estimated height by the VSI method. Fig. 4 is an explanatory diagram for obtaining the estimated height by the VSI method. Fig. 5 is a processing flowchart of the method for measuring the surface shape of the first embodiment -12- 565681 V. Description of the invention (11) Fig. 6 is a block diagram of the main structure of the surface shape measuring device of the second embodiment. [Best Mode for Invention] The following is a mode for solving conventional problems. < First Embodiment > A first embodiment of the present invention will be described with reference to the drawings. Fig. 1 is a block diagram showing the structure of a surface shape measuring device according to the first embodiment of the present invention, Fig. 2 is an explanatory diagram illustrating a mechanism of interference fringes, and Figs. 3 and 4 are diagrams for obtaining an estimation by the VSI method An explanatory diagram of the height, and FIG. 5 is a processing flowchart of the surface shape measuring method of the first embodiment. As shown in FIG. 2, the distance from the semi-transparent mirror 16 to the control surface 15 is L1, and the plane at a distance from the semi-transparent mirror I 6 by L1 is the plane E. The height h from that point to the plane E is the position of the interferometer, and the height of the point P on the measurement target surface 31 of the sample is hP. The surface shape measuring device of the first embodiment As shown in FIG. 1, the structure includes an optical system element 1 for irradiating a fine pattern formed on a measurement object 30 such as a semiconductor wafer, a glass substrate, or a metal substrate with white light in a specific frequency band, and control for controlling the optical system element 1.系 Element2。 System element 2. The structure of the optical system element 1 includes: a white light source 10, which generates white light which is irradiated on the measurement target surface 31, and a reference surface 15; a collimating lens 11, which causes the white light from the white light source 10 to be parallel light; Bandpass filter [2], white light with a narrower frequency band than white light from the white light source 10 (in the i--13-565681 V. Description of the invention (12) In the embodiment, the center wavelength is 600nm and the bandwidth is 20nm); half The transparent mirror 13 reflects white light that has passed through the band-pass filter 12 or white light that has been directly irradiated from the white light source 10 without passing through the band-pass filter 12 toward the measurement object 30, and allows the light from the measurement The white light in the direction of the object 30 passes; the objective lens 14 collects the white light reflected by the semi-transparent lens 13; the semi-transparent lens 16 divides the white light passing through the objective lens 14 into a control reflected on the control surface 15 The light and the measurement light passing through the measurement target surface 31 are simultaneously concentrated again on the reference light reflected on the reference surface 15 and the measurement light reflected on the measurement target surface 31 to generate interference fringes; the actuator 17 can Perform objective lens 1 4. Control surface 15 and semi-transparent mirror 16 are driven up and down and left and right; imaging lens 18 is used to form an image of white light formed by collating control light and measurement light; and CCD camera 19 is used to capture interference fringes and the measurement target surface The image of 31. The white light source 10 may be, for example, a white light lamp, which generates white light with a wide frequency band. The white light generated by the white light source 10 becomes parallel light due to the collimator lens 11 and is incident on the band-pass filter 12. The band-pass filter 12 is a filter that allows only white light in a specific frequency band to pass, and is installed on the light path from the white light source 10 to the CCD camera 19. However, the installation position is preferably located on the optical path between the white light source 10 and the white light source 10 divided into the control light of the control surface 15 and the measurement light of the measurement target surface 31. In this embodiment, it is installed on the optical path between the collimator lens 11 and the translucent mirror 13. For the band pass filter 12, the band-pass optical interference filter with a center wavelength of 600 nm and a bandwidth of 20 nm is used. -14-565681 V. Description of the invention (13). The white light of the wider frequency band incident on the band-pass filter 12 passes through the band-pass filter 12 only of the white light of a specific frequency band with a narrower frequency band. The band-pass filter 12 corresponds to the band limiting mechanism of the present invention. In this embodiment, the band-pass filter 12 can be switched by the filter switching unit 24 described below. In steps S 1 to S5 described later, the band-pass filter 12 can be deactivated by the filter switching unit 24. Except for the optical path between the collimator lens 11 and the translucent mirror 13, the white light generated by the white light source 10 can maintain a wide frequency band without passing through the band-pass filter 12. After step S6 described below, the filter switching unit 24 can be used to position the band-pass filter 12 on the optical path between the collimator lens 11 and the translucent mirror 13, and the white light generated by the white light source 10 passes through the band pass. The filter 12 is limited to white light in a narrower frequency band. In this specification, white light that is directly irradiated by the white light source 10 without passing through the band-pass filter 12 is defined as "the first white light." At the same time, white light that passes through a narrower band of the band-pass filter 12 is defined. Defined as "2nd white light". The semi-transparent mirror 13 can reflect the first white light or the second white light in the direction of the measurement object 30 and allow the white light returned from the direction of the measurement object 30 to pass. Here, the first and second white light reflected by the translucent mirror 13 are incident on the objective lens 14. The objective lens 14 is a lens that collects incident white light in the direction of the focal point p. The first / second white light collected by this objective lens 14 passes through the control surface 15 and reaches the translucent lens 16. -15- 565681 V. Description of the invention (14) The semi-transparent lens 16 divides the first / second white light collected by the objective lens 14 into a contrast for the purpose of reflecting on the contrast surface 15 and reflecting on the semi-transparent lens 16 At the same time, the light and the measurement light that has passed through the translucent mirror 16 for the purpose of reflecting on the measurement target surface 31, and at the same time, the control light and the measurement light are integrated again to generate interference fringes. The white light reaching the semi-transparent mirror 16 is divided into a reference light reflected by the semi-transparent mirror 16 and a measurement light passing through the semi-transparent mirror 16. The control light reaches the reference surface 15 and the measurement light reaches the measurement. Target surface 3] 〇 The control light reflected on the control surface 15 will reach the semi-transparent mirror 16, and then, the control light will be reflected again on the semi-transparent mirror 16. The measurement light passing through the translucent mirror 16 collects light toward the focal point P and reflects it on the measurement target surface 31. This reflected measurement light reaches the semi-transparent mirror 16 and passes through the semi-transparent mirror 16. The semi-transparent mirror 16 will collect control light and measurement light again. At this time, the optical path differs depending on the distance L1 between the reference surface 15 and the semi-transparent mirror 16 and the distance L2 between the semi-transparent mirror 16 and the measurement target surface 31. The control light and the measurement light will interfere with each other corresponding to this optical path difference, and generate interference fringes. The white light in the state where the interfering fringes are generated passes through the semi-transparent mirror 13 and an image is formed by the imaging lens 18, and then enters the CCD camera 19. The actuator 17 is for changing the distance difference between the fixed distance L1 between the reference surface 15 and the semi-transparent mirror 16 in the optical system element 1 and the variable distance L2 between the semi-transparent mirror 16 and the measurement target surface 31. The objective lens 14, the control surface-16-565681 V. Description of the invention (15) 15, and the translucent lens 16 can be moved in the direction of the three axes that intersect perpendicularly. The structure of the device is based on the instructions of the CPU 20 in X, The objective lens 14, the control surface 15, and the translucent lens 16 are driven on the γ and Z axes. In this embodiment, the objective lens 14, the reference surface 15, and the translucent lens 16 are respectively moved. However, the optical element 1 may also be moved as a whole, for example, a work on which the measurement object 30 is placed. The stage (not shown) moves in 3 vertical directions. The actuator 17 corresponds to the changing mechanism of the present invention. In addition, the variable mechanism of the present invention may be driven by a driving mechanism such as a three-axis drive type servo motor to perform vertical, horizontal, and left-right driving. The CCD camera 19 captures white light in a state where interference fringes are generated, and an image that is reflected by the measurement light and is located near the focal point P of the measurement target surface 31. The control system component 2 collects the captured image data. In addition, the objective lens 14, the reference surface 15, and the translucent lens 16 are moved up, down, left, and right by the actuator 17. The distance between the distance L1 and the distance L2 can be changed by using the above-mentioned movement in the up-down direction. In this way, the interference fringes can be changed according to the distance difference between the distance L1 and the distance L2. At each sampling interval described later (λ / 8 in the i-th embodiment), changes in interference fringes and images of the measurement target surface 31 are captured by the CCD camera 19, and the image data is collected by the control system element 2. The CCD camera 19 corresponds to the photographing mechanism of the present invention. The sampling interval corresponds to a specific interval in the present invention. The control system component 2 comprehensively controls the overall surface shape measuring device. In its structure, it has: CPU20, which performs specific calculation processing; remembers the body 21, stores the image data gradually collected by CPU20, and the calculation of CPU20 1 L --- 565681 V. Description of the invention (16) Various data such as results; input unit 22 'enters the sampling interval or other setting information, such as mouse and keyboard; monitor 23' displays the measurement target surface 3 丨 image, etc .; filtering The device switching unit 24 switches the computer system of the band-pass filter 12 according to the instruction of cpU20. The CPU 20 corresponds to the sampling mechanism and the calculation mechanism of the present invention. CPU20 is also called a central processing device. In addition to controlling the CCD camera 19, the memory 21, and the actuator 17 of the optical system 1, the CPU 20 is based on the measurement target surface 31 which contains the interference fringe captured by the CCD camera 19. The image data is processed to measure the uneven shape of the measurement target surface 31. This process will be described in detail later. In addition, the CPU 20 is connected to the monitor 23 and an input unit 22 such as a keyboard or a mouse, and the operator can input various setting information from the input unit 22 while observing the operation screen displayed on the monitor 23. After the measurement of the measurement target surface 31 is completed, the monitor 23 displays the uneven shape of the measurement target surface 31 as a number or an image. In steps S1 to S5 to be described later, the filter switching unit 24 may directly cause the white light generated by the white light source 10 to enter the measurement object 30 without passing through the band-pass filter 12, that is, directly irradiate the first white light to the measurement target 30. On the measurement object 30, the band-pass filter 12 is moved out of the optical path between the collimator lens π and the translucent mirror 13, and after the following step S6, in order to make only a white light source with a narrow frequency band The white light generated by 10 can pass through the band-pass filter 12 and be incident on the measurement object 30, so that the band-pass filter 2 is located in the optical path between the collimating lens Π and the translucent mirror 13. That is, the filter switching unit 24 is a device that allows the band-pass filter 12 to move inside and outside the optical path between the collimator lens 11 and the translucent mirror i 3 -18- 565681. 5. Description of the invention (17). Similar to the CCD camera 19, the filter switching section 24 is structured to drive the band-pass filters 12 on the X, Y, and Z axes in accordance with instructions from the CPU 20. Hereinafter, the processing of the entire surface shape measuring apparatus according to this embodiment will be described with reference to the flowchart in FIG. 5. Step S1 (set as the first white light) First, when the filter switching unit 24 is in a state where the band-pass filter 12 is located outside the optical path between the collimator lens 11 and the translucent mirror 13, the white light source 10 generates White light. Therefore, the white light generated by the white light source 10 will maintain a wide frequency band without passing through the band-pass filter 12, and will be incident on the measurement object 30 and the control surface 15. Since the band-pass filter 12 is located outside the optical path between the collimator lens 11 and the translucent mirror 13, the first white light setting is implemented. From the setting to the second white light setting in step S6, the measurement target surface 31 and the reference surface 15 are irradiated with the first white light. Step S2 (Variation of the relative distance at a specific interval) The CPU 20 provides an instruction to start movement of the actuator 17, so that the optical system element 1 originally located at a specific measurement position starts to move in the Z-axis direction. Actuator 17 causes objective lens 14, control surface 15, and translucent lens 16 to move only a preset distance in the Z-axis direction. Therefore, the relative distance between the reference surface 15 and the measurement target surface 31 varies. This step S2 corresponds to the first step of the present invention. In this embodiment, the relative distance is changed according to a specific interval of 1/8 (λ / 8) of the wavelength λ, that is, the sampling interval, to move it in the direction of the Z axis. -19-565681 V. Description of the invention ( 18). Also, when the initial sampling position is h0 and the sampling interval is △, the sampling intervals in the z-axis direction are + △, hG + 2 △, ..., 1 !. + (11-1) Move up and mark the image data of the measurement target surface 31 including the interference fringes captured by the CCD camera 19 at each sampling interval. As shown in Figure 3, i is attached to the sampling position. The sample number is h, and the sample number is attached when △. Subsequent numbers are attached accordingly. In addition, if the interference light intensity (interference pattern) reflected from the two planes of the reference surface 15 and the measurement target surface 31 is g (h), each sampling interval is g (hQ), g (hQ + A), g (hQ + 2A), ..., g (hQ + nA), and obtain the interference light intensity unitary group. Step S3 (Is it possible to obtain a specific number of data?) If a specific number of data cannot be obtained from the interference light intensity group, return to step S2, and in step S2 change the relative distance at a specific interval (sampling interval), and at each sampling interval The control surface 15 and the measurement target surface 31 were irradiated with white light to obtain interference light intensity. If a specific number of data can be obtained, the process proceeds to step S4. This step S3 corresponds to the second step of the present invention. Step S4 (obtaining the envelope curve) In the interference light intensity group, the sampling position where the interference light intensity changes the most is the height of the measurement target surface. However, in fact, the sampling data at each specific interval will be scattered, so it is not easy to change from g (h) Find the position where the interference light intensity is the maximum. Therefore, the height of the measurement target surface is measured based on the envelope curve of the interferogram or the position where the characteristic eigenvalue of the invention disclosed in "Japanese Patent Application Laid-Open No. 2001-0661 22" previously proposed by the inventor is the largest. -20 · 565681 V. Description of the invention (19) In this embodiment, the envelope curve is obtained and the position at which the envelope curve is maximized is taken as the estimated height, and the estimated height is h. Subtract the average 値 of the interference light intensity 値 group from each interference light intensity 値 group of the envelope curve obtained, and then square the data to be reduced. Therefore, the envelope curve shown in Fig. 4 can be obtained by smoothing the band-pass filter 12. Step S5 (obtaining the estimated height) After obtaining the envelope curve, obtain the estimated height h where the envelope curve is the largest. If the image number at this time is i, the height h is estimated and becomes b + J △. This step S5 corresponds to the third step of the present invention, and also corresponds to the first process of the present invention. Step S6 (set as the second white light) The filter switching unit 24 positions the band-pass filter 12 on the optical path between the collimator lens 11 and the translucent mirror 13. The white light generated by the white light source 10 passes through the band-pass filter 12 and the white light restricted by the narrower frequency band is incident on the measurement object 30 and the reference surface 15. Since the band-pass filter 12 is located on the optical path between the collimator lens 11 and the translucent mirror 13, the second white light is set. From the setting to the display of step SU, the measurement target surface 31 and the reference surface 15 are irradiated with the second white light. Step S7 (precipitation of sampling positions close to the estimated height) Four sampling positions close to the estimated height are deposited. In this embodiment, 'h, 1 ^ + Δ, M + 2Δ, h, + 3 △ are precipitated, however, it can also be hi, hi + Δ, h, + 2 △, or 1 ^ -2 ^, 1m "Δ, hi, h, △, or h2 3 Δ'h, -2A, 1 ^ · Δ ,. -21-565681 V. Description of the invention (2) and 'The sampling interval △ is λ / 8, interference The amount of change in the light phase corresponds to π / 2. Then, it is set at the precipitation position and proceeds to step S8. Step S8 (obtaining the interference light phase) h, hi + Δ, 1 ^ + 2Δ, h + 3 △ The intensity of each interference light is A, B, C, D. Moreover, the phase difference of the interference light that must be obtained is 4, and the phase difference 0 can be obtained by using A, B, C, and D according to the formula (2). Step S8 corresponds to the fourth step of the present invention, and also corresponds to the second process of the present invention. Step S9 (to determine the height of the measurement target surface) The height of the measurement target surface finally obtained is h2. If phase 4 is taken, Using h on the left side of the formula (1) as h2, choose the integer N that makes the height h2 closest to the estimated height h, and then obtain the height h2 of the measurement target surface from the selected N. This step S9 is equivalent to the fifth of the present invention Step, also corresponds to the third process of the present invention Step S10 (End all specific positions?) Until all specific positions are completed, the CPU 20 repeats the processing of steps S 1 to S9 to obtain the heights of all the specific positions. In addition, a television camera (2-dimensional CCD) is actually used. Detector), so step S 10 will only repeat the calculation steps. Step S11 (display) The CPU 20 will display the height information of the specific position on the monitor 23, and display the 3D or 2 based on the height information of each specific position. Dimensional images. The operator can observe these displays and grasp the uneven shape of the measurement target surface 31 of the measurement target 30. -22- 565681 V. Description of the invention (21) According to the first embodiment, a specific interval will be provided in step S2. The sampling interval △ changes the relative distance between the reference surface and the aforementioned measurement target surface. In step S3, each time the sampling interval △ changes, the intensity of the interference light of the white light on the reference surface and the measurement target surface is measured. In step S5, the estimated height at which the interference light intensity is the largest among the interference light intensity groups measured in step 3 is obtained. In this embodiment, in step S, 4. Obtain the envelope curve, and in step S5, obtain the maximum estimated height of the envelope curve. From step S1 to step S5, until the interference light intensity is measured, it is a step of the vertical scanning interference method (VSI method). In S8, the phase difference 4 of the interference light is obtained based on the estimated height estimated in step S5. In step S9, the measurement target surface is obtained based on the estimated height estimated in step S5 and the phase difference 4 of the interference light obtained in step S8. In step S9, the phase of the interference light is obtained according to the intensity of each interference light, and the phase shift interference method (PSI method) is a step until the height of the measurement target surface is measured based on the phase measurement. Therefore, the height of the measurement target surface can be determined according to the estimated height estimated in step S5, without performing un-lapping processing for the purpose of obtaining the height of the measurement target surface, and not only can the measurement target surface be a smooth surface Or, it can also measure the surface with large surface segment difference. In addition, since the estimated height is obtained in step S5, and the height of the measurement target surface is obtained in step S9, it is possible to perform a highly accurate measurement. As a result, a measurement target surface having a large surface segment difference can also be measured with high accuracy. When the sampling interval △ is λ / 8/8, the phase of each interference light at the four sampling positions changes -23-565681 V. Description of the invention (22) It is π / 2 corresponding to λ / 8/8, and in step S8, according to each sampling interval The interference light phase of △ is used to determine the interference light phase. At this time, it is necessary to move the phase variation of the interference light by π / 2 between 0, π / 2, π, and 3π / 2. Since λ / 8 corresponding to the phase variation can be directly used, it is not necessary to re-implement the waveform reduction. The interference light phase is obtained by using the interference light intensity of each of the sampling intervals λ / 8. In addition, the light irradiated on the control surface and the measurement target surface is white light. Compared with monochromatic light, the interference of white light is lower. The maximum interference light intensity change (sharpness) is obvious. In step S5, it is easier to obtain an estimated height at which the intensity of the interference light is the maximum. In addition, step S3 is part of the steps of the so-called VSI method, and after step S 8 is part of the steps of the so-called PSI method, so the white light (the first white light) irradiated by steps S1 to S6 is not directly used. Steps S7 to S10 are white light (second white light) whose irradiation frequency band is narrower than the white light irradiated in steps S1 to S6, and the interference light phase is obtained according to the intensity of each interference light related to the second white light. In order to obtain the phase in step S8, the wavelength λ must be known. Therefore, this optical system has a narrower band of white light than the first white light irradiated in step S3. Moreover, in this embodiment, the wavelength of the second white light cannot be completely determined due to the non-monochromatic light. However, since the frequency band of the second white light is relatively narrow, a specific wavelength in the frequency band can be viewed. It is a known wavelength λ, and the sampling interval Δ which determines λ / 8 may also be used. < Second embodiment > Next, a second embodiment of the present invention will be described. The same parts as in the first embodiment -24- 565681 V. Description of the Invention (23) The same symbol is attached and the description is omitted. As shown in FIG. 6, the surface shape measuring device of the second embodiment has “the optical system element 1 has a white light source 10b different from the white light source 10 and the control system element 2 has a switching white light source 10, 10b. The light source switching section 25 replaces the filter switching section 24 in the control system element 2 of the first embodiment. In terms of structure, the frequency band of the white light generated by the white light source 10b is narrower than that of the white light generated by the white light source 10. That is, the white light generated by the white light source 10 is the first white light, and the white light generated by the white light source 10b is the second white light. Since the processing of the surface shape measuring method is the same as that of the first embodiment, its description is omitted. The difference from the first embodiment is that there is no band-pass filter 12 for limiting white light to a narrower frequency band. However, it is preferable to have a band-pass filter 12 ° The present invention is not limited to the foregoing embodiment, It can be modified as shown below. (1) In the foregoing first and second embodiments, the optical system element 1 is a so-called "Mikeson interferometer". However, an optical system element called "(? P22L21-22) MIRAN interferometer" ( Refer to "Japanese Patent Application Laid-Open No. 200 1 -066 1 22"). (2) In the aforementioned first and second embodiments, the phase is obtained by the so-called "4-point data method" by precipitating 4 points of hl, hjA, 1 ^ + 2Δ, h! + 3 △. However, as long as there is Three or more relevant data on the intensity of the interference light may be used, and the representative is, for example, the 3-point data method and the 5-point data method. -25-565681 V. Description of the invention (24) (3) In the first and second embodiments, the VSI method uses white light with a wide frequency band (the first white light) 'and the PSI method uses white with a narrow frequency band. Light (second white light), however, if the wavelength; ^ can be limited to some extent, the PSI method can also use the same white light as the VSI method. In another example, the PSI method is white light in a narrow frequency band, however, it may be monochromatic light. In the case of monochromatic light, since the wavelength λ is completely known, it is easier to determine the sampling interval Δ. In the first embodiment, the band-pass filter and the wave filter 12 are filters that can pass only monochromatic light. In the second embodiment, in addition to the white light source 10, 尙 has a monochrome Light source. (4) In the example described above, the sampling interval in the first embodiment is; (/ 8, however, it can be larger. When waveform reduction is not implemented, and N is an integer, the sampling interval can also be (N / 2 + 1/8) × Again. Since N λ / 2 corresponds to 1 per week, so as long as this sampling interval corresponds to the phase, sampling in different periods can be implemented. That is, when N is 1, as long as it corresponds to 0, 5 λ / 8 (= λ / 2 + λ / 8), 5 Jiu / 4 (= λ + λ / 4), ..., !!] The interference light phase will be at 0, 5π / 2 (= 2π + π / 2), 5π (= 4π + π),… moves. Of course, N can also be other than 1, and the N of each sample also changes randomly. At this time, the phase of the interference light will also be at intervals of π / 2 at 0, π / 2, π, 3π / 2 changes, so you can directly use λ / 8 corresponding to the phase change without re-implementing the waveform. When restoring the waveform, there are methods such as the following. When using the sampling theorem, the sampling interval is only greater than λ / 8. It is only necessary to be smaller than the Nyquist interval. When using the bandpass sampling theorem, the sampling interval is only required to be greater than the Nyquis interval. No matter -26-565681 V. Description of the invention (25) In any case, use sampling Although the theorem or the band-pass theorem need to implement waveform reduction, because the sampling interval is greater than λ / 8, compared with the sampling interval of λ / 8, it only needs to obtain fewer samples of the interference light intensity, which can shorten the interference light. Sampling time of the intensity. (5) In the aforementioned first embodiment, the envelope curve is used as the intensity of the interference light. However, an interference pattern may also be used. In addition, "Japanese Patent Application Laid-Open No. 2001-06-06" previously proposed by the inventor may also be used. [22] The characteristic culvert of the invention is regarded as the intensity of the interference light, and then the position where the characteristic culvert is the maximum can be obtained. (6) In the aforementioned first and second embodiments, the VSI method and the PSI method use different lights. Therefore, it is better to change the relative distance (L1 + L2) of the reference surface and the measurement target surface to obtain the interference light intensity in step S2, and to change the relative distance of the reference surface and the measurement target surface to obtain the interference light intensity again in step S6. Of course, when the same light as the VSI method is used in the PSI method, it is not necessary to perform another change to obtain the interference light intensity. [Industrial Application Possibility] As described above, the surface shape measuring method of the present invention The device and its device use optical interference to measure the uneven shape of the measurement object, and it is not only suitable for precision processed products such as semiconductor wafers or glass substrates for liquid crystal displays, but also for semiconductor bumps or metal processed surfaces with large surface segment differences. Measurement target surface. [Explanation of component symbols] 1 Optical system component 2 Control system component -27- 565681 V. Description of the invention (26) 10, 10 b White light source 11 Collimation lens 12 Bandpass filter 13 > 16 Half Transparent lens 14 Objective lens 15 Control surface 17 Actuator 18 Imaging lens 19 CCD camera 20 CPU 21 Memory 22 Input section 23 Monitor 24 Filter switching section 25 Light source switching section 30 Measurement target 31 Measurement target surface 50 Sample table-28-