201144749 六、發明說明： 【發明所屬之技術領域】 本發明係有關於一種量測孔洞深度之方法，尤指一種 利用跨焦影像的方式擷取孔洞不同區域之至少二個^像， 並以聚焦演算法（FQeai Metrie)分別分析不同影像的聚焦 位置’可提高量測精確度，尤適用於量測高深寬 洞 深度之方法。 【先前技術】 傳統上晶片是製作在二維⑽空間上的，然而隨著晶 片複雜度的增加，使得在二維空間上的橫向面積加大，已 無法讓摩爾定律（Μ_，s㈣能繼續有效。因此，逐漸 有亡考慮到利用堆疊的方式來整合不同的晶片，因而創造 出三維晶片（3D IC)。三維晶片製作上利用矽穿孔 (Jhrough-Sillcon Via)封裝技術，透過以垂直導通來整合 日日圓隹且的方式，以達到晶片間的電氣互連，因此石夕穿孔 的製作的好壞及量測上的精確度將明顯的影響晶片的良 率、。石夕穿孔在製作上為—高深寬比的孔洞，而且為一盲孔， 無法直接以光學顯微鏡量測其深度。 …種畺测岗深寬比的孔洞深度的方法，其係利用 光學置測系、統量測孔洞之不同深度位置，並擷取每-深度 位置之'5V像’利用聚焦演算法（F〇cal臉士）計算並紀錄 聚…、位置#選取聚焦演算數值的最大值進行計算，請參 閱第一圖所示聚焦位置曲線圖，於該曲線呈現第一組及第 201144749 二組兩組波峰，第一組波峰位於180〜21〇/zm範圍内，代表 孔洞之上孔（孔洞開口處）聚焦位置，第二組波峰位於 100〜12()/ini範圍内，代表孔洞之下孔（孔洞底部）聚隹位 置，經由軟體選取各組波峰之最高聚焦演算數值，亦即'影 像之聚焦位置，其中，上孔聚焦位置ρι約為188㈣，； 孔聚焦位置P2約lG8//m ’將兩聚焦位置相減後所得數值 即為孔洞的深度’ ，因此，孔洞深度 為80//1H’由第-圖所示曲線可知，上述習知量測孔洞^ 度方式之缺點在於’由於下孔之聚焦演算數值差異不明 顯，無法正確取得下孔聚焦位置，導致所計算之孔洞深度 誤差大’ &應用於晶片石夕穿孔製作時，必會明顯影響晶 的良率。 就習知揭露量測孔洞深度之專利而言，例如美國專利 =⑽2436 號「Non-c〇ntact h〇le depth _」，請參閱 第一圖所不其量測系統之架構示意圖，其主要包括一昭明 ;影像㈣單元2()、—待測工件ig以及待測工 方式包括以下步驟： 里、』孔冰之 魅^ ^ :量測下孔直徑；使用影像制單^ 2〇量測 °哀埋頭孔16之下孔直徑（影像）。 程式孔::下孔直徑;使用控制單元130的分析 旦計算上孔距離；即第二圖中該卫件ig表面至 的位署7單=20之該上孔距離160 ’由於影像價測單元20 疋固定的，故該上孔距離160可以很容易得知。 201144749 ^步驟308計算下孔距離；即第二圖中該下孔16底緣至 衫像偵測單元20之該下孔距離164，由於不同的孔深其在 影像偵測單元20上所量測到的下孔直徑是不同的，即所量 測的下孔直徑與孔深是相關的，故可由所量測的下孔直徑 得出該下孔距離164。 工 步驟310計算孔深；由步驟3〇6及步驟3〇8所得知的 上孔距離160及下孔距離164可以計算出埋頭孔16的深度 並不適用於高 准δ亥案所提供之方式僅適用於穿透孔 深寬比的盲孔。 【發明内容】 有鑑於習知技術之缺失，本發明提出一種量測孔洞深 又之方法，利用跨焦影像的方式擷取孔洞不同深度位置的 影像，並以聚焦演算法㈤—分別分析上孔及下 孔的聚焦位置’可提高量測精確度，尤適用於量測高深 比的孔洞深度。 為達到上述目的，本發明提出一種量測孔洞深度之方 法，包含： 、光予量測系統擷取一孔洞不同區域之至少二個影 像； 以- 括.决算法計算一該影像，以得到複數個第一演算數 值及八對應的複數個第一位置數值； H貝异法計算另一該影像’以得到複數個第二演算 201144749 數值及其對應的複數個第二位置數值；及 決疋該複數個第一演算數值中之最大值及其第一位置 數值，及決定該複數個第二演算數值中之最大值及其第二 位置數值’其中第—位置數值與第二位置數值之差值即為 該孔洞之深度。 為使貴審查委員對於本發明之結構目的和功效有更 進一步之了解與認同，茲配合圖示詳細說明如后。 【實施方式】 以下將參照隨附之圖式來描述本發明為達成目的所使 _$術手段與功效，而以下圖式所列舉之實施例僅為辅 助說明’以利f審查委員瞭解，但本案之技術手段並不限 於所列舉圖式。 '彡閱第二圖所不’本發明所提供之量測孔洞深度之 係利用—光學量測系統5 〇量測-物件6 〇之孔洞6 i， 孔ϋ具有—深度H ’且孔洞61為盲孔光學量測系統 二二Γ:孔洞61之深度H方向往復移動，藉由光學量 1測孔洞61不同區域，亦即不㈣度位置，並 置之孔洞61之至少二個影像，如此，可得 =1Γ像，使用者可自行設定掏取影像之次數， 次數越多’對於計算孔洞61深度Η之精確度 明的是，—般孔洞61之斷面呈圓形，因此 擷==現圓環狀，若孔洞61斷面為_形、矩 ;他幾=狀，則所掏取的影像也會相對應 王現橢_、矩形、多邊形或其他幾何形狀之環狀。 201144749 當光學量測系統50完成上述擷取影像之步驟後，再利 用聚焦演算法根據所擷取之環狀影像進行分析及演算，一 般常用的聚焦演算法（Focal Metric)包括梯度能量法 (Gradient Energy) ’ 標準差法（Standard deviation)，拉 普拉斯法（Laplacian)及對比法（Contrast)。 以gi(x，y)表示在影像i的位置（x，y)上的強度。梯度 能量法（Gradient Energy)的演算法如下： = Σ Σ ((-s'v *(·Λ^ ·ν，))2 + {^· *(a% i·))2 ) .V V · 其中， ' · <?, (·Υ〇.) = T(—見(·、· 一 L ·ν — 1) + g,·(Λ· +1· ·ν — i) 2ι?ί(Λ·-1,ν) + 2^(Λ·+Ιν)--(Λ-1,ν+1)+^(Λ. + 1^+1))201144749 VI. Description of the Invention: [Technical Field] The present invention relates to a method for measuring the depth of a hole, in particular, a method for capturing at least two images of different regions of a hole by means of a transfocal image, and focusing The algorithm (FQeai Metrie) analyzes the focus position of different images separately to improve the measurement accuracy, especially for measuring the depth of high and deep holes. [Prior Art] Traditionally, wafers have been fabricated in two-dimensional (10) space. However, as the complexity of the wafer increases, the lateral area in two-dimensional space increases, and Moore's Law (Μ_, s(4) can no longer be effective. Therefore, the gradual decline considers the use of stacking to integrate different wafers, thus creating a three-dimensional wafer (3D IC). The three-dimensional wafer fabrication uses Jhrough-Sillcon Via packaging technology to integrate through vertical conduction. The way of the Japanese yen is to achieve the electrical interconnection between the wafers. Therefore, the quality of the production and the accuracy of the measurement will significantly affect the yield of the wafer. A hole with a high aspect ratio, and a blind hole, cannot directly measure its depth by optical microscopy. The method of measuring the depth of the hole with an aspect ratio is based on the difference between the optical measurement system and the measurement hole. Depth position, and extract the '5V image' of each depth position using the focus algorithm (F〇cal face) to calculate and record the poly..., position # select the maximum value of the focus calculation value For calculation, please refer to the focus position curve shown in the first figure. The curve shows the first group and the 201144749 two sets of two peaks. The first set of peaks is in the range of 180~21〇/zm, representing the hole above the hole ( At the opening of the hole, the focus position, the second set of peaks is in the range of 100~12()/ini, which represents the position of the hole below the hole (bottom of the hole), and the highest focus calculation value of each group of peaks is selected via software, that is, ' The focus position of the image, wherein the focus position of the upper hole ρι is about 188 (four), and the focus position of the hole P2 is about lG8//m 'the value obtained by subtracting the two focus positions is the depth of the hole', and therefore, the hole depth is 80/ /1H' can be seen from the curve shown in the first figure. The disadvantage of the above-mentioned conventional measurement method is that 'the difference in the focus calculation value of the lower hole is not obvious, and the focus position of the lower hole cannot be obtained correctly, resulting in the calculated hole depth. When the error is large, it will obviously affect the yield of the crystal when it is applied to the wafer. For the patents that disclose the depth of the hole, for example, US Patent = (10) 2436 "Non-c〇ntact H〇le depth _", please refer to the schematic diagram of the measurement system of the first figure, which mainly includes a schematic; image (4) unit 2 (), - workpiece ig to be tested and the method to be tested include the following steps: , "The charm of Kong Bing ^ ^ : Measure the diameter of the lower hole; use the image to make a single ^ 2 〇 measurement ° bury the hole diameter under the hole 16 (image). Program hole:: the diameter of the lower hole; using the control unit 130 The upper hole distance is calculated by the analysis; that is, the distance of the upper hole of the guard ig surface in the second figure to the single 7=20 is fixed by the image measuring unit 20 疋, so the upper hole distance 160 can be very It is easy to know. 201144749 ^Step 308 calculates the lower hole distance; that is, the bottom hole edge of the lower hole 16 in the second figure to the lower hole distance 164 of the shirt image detecting unit 20, because the different hole depth is in the image detecting unit 20 The diameter of the lower hole measured is different, that is, the measured diameter of the lower hole is related to the depth of the hole, so the distance 164 of the lower hole can be obtained from the measured diameter of the lower hole. Step 310 calculates the hole depth; the upper hole distance 160 and the lower hole distance 164 known from steps 3〇6 and 3〇8 can be used to calculate the depth of the countersunk hole 16 and are not suitable for the manner provided by the Micro Motion δ Hai case. Only for blind holes that penetrate the hole aspect ratio. SUMMARY OF THE INVENTION In view of the lack of the prior art, the present invention proposes a method for measuring the depth of holes, using the method of transfocal images to capture images of different depth positions of the holes, and analyzing the upper holes by focusing algorithm (5). And the focus position of the lower hole can improve the measurement accuracy, especially for measuring the depth of the hole in the aspect ratio. In order to achieve the above object, the present invention provides a method for measuring the depth of a hole, comprising: a light pre-measurement system capturing at least two images of different regions of a hole; and calculating an image by a method to obtain a complex number a first calculus value and a plurality of first position values corresponding to eight; the H-shell method calculates another image 'to obtain a plurality of second calculus 201144749 values and corresponding plurality of second position values; and a maximum value of the plurality of first calculus values and a value of the first position thereof, and a maximum value of the plurality of second calculus values and a difference between the second position value and the second position value That is the depth of the hole. In order to enable the reviewing committee to have a better understanding and approval of the structural purpose and efficacy of the present invention, the detailed description is as follows. [Embodiment] Hereinafter, the present invention will be described with reference to the accompanying drawings, and the embodiments shown in the following drawings are merely for the purpose of explanation. The technical means of this case are not limited to the illustrated figures. The measurement of the depth of the hole provided by the present invention is measured by the optical measuring system 5 〇 - the hole 6 i of the object 6 ,, the hole has a depth H ' and the hole 61 is The blind hole optical measuring system is two-dimensionally: the depth of the hole 61 is reciprocated in the H direction, and the optical quantity 1 is used to measure different areas of the hole 61, that is, not at the (four) degree position, and at least two images of the hole 61 are juxtaposed. If you get 1 Γ image, the user can set the number of times the image is captured. The more the number of times, the better the accuracy of calculating the depth of the hole 61 is, the section of the hole 61 is circular, so 撷 == now Ring shape, if the hole 61 has a _ shape and a moment; if it is a shape, the captured image will also correspond to the ring shape of the ellipsoid, rectangle, polygon or other geometric shape. 201144749 After the optical measurement system 50 completes the above steps of capturing images, the focus algorithm is used to analyze and calculate according to the captured ring image. The commonly used focus algorithm (Focal Metric) includes gradient energy method (Gradient) Energy) 'Standard deviation, Laplacian and Contrast. The intensity at the position (x, y) of the image i is represented by gi(x, y). The algorithm of Gradient Energy is as follows: = Σ Σ ((-s'v *(·Λ^ ·ν,))2 + {^· *(a% i·))2 ) .VV · where , ' · <?, (·Υ〇.) = T (- see (·, · a L · ν - 1) + g, · (Λ· +1· · ν — i) 2ι?ί(Λ· -1,ν) + 2^(Λ·+Ιν)--(Λ-1,ν+1)+^(Λ. + 1^+1))
標準差法（Standard deviation)的演算法如 FMStandard deviation algorithm such as FM
n-\ 下： 拉普拉斯法（Laplacian)的演算法如下： FM= H(1、、. + Ζ”） 其中， 4.+-沙-l'v -1) _ 4g· (·Υ - l'v) - g (·'. 乂V+Ρ -4g, (.v -1, ν -1)+20g·. (,ν - L v) -4^(. (Λ· -1, r+^ (·、· 一 一 i) 一％ (·γ_1· .r)-总(Λ·-υ'+ι) 201144749 對比法（Contrast)的演算法如下 imn F\4 α Connasr min 上述聚焦演算方法為一般習知的方法，其詳細的演算 内容不予詳述。本發明的特點在於利用聚焦演算法針對所 擷取之環狀影像之一周緣區域及一内側區域進行運算。請 參閱第四圖所示’說明本發明定義環狀影像之周緣區域之 方式:其顯示所擷取之一圓環狀影像70 ’該周緣區域8〇 為-等寬之圓形環狀區域（第四圖示斜格線區域），該圓形 =狀贫區域具有H側周緣81 m —第—外側周緣 82’第-内側周緣81與圓環狀影像7〇具有一第一距離di， 该第-内側周緣81與該第—外側周緣82之間具有一寬产 W’該周緣區域80係涵蓋該孔洞61見緣，較理想之狀況是， 該圓形環狀影像70係位於該第—内侧周緣81與該第一外 側周緣82中間（如第四圖所示），但並不限於此，此外，即 =擷取之孔洞影像為圓環狀，本發明該周緣區域也不限 ::圓環狀’以第四圖而言，第-内側周緣81為圓形，第 妝，:周ί 82則可為糖圓形、矩形、多邊形或任意幾何形 之’本發明所定義之周緣區域，其設定原則在 d/,、日:内側周緣81與圓環狀影像7Q纟有—第一距離 尺+廿=周緣區域80可涵蓋該圓環狀影像7〇即可，形狀 尺寸並無一定限制。 區域第五圖所示’說明本發明定義環狀影像之内侧 S 〆本發明所定義之内侧區域90(第五圖示斜格 201144749 線區域）係位於該圓環狀影像7以，該内側 心’但不包括該細周緣，内側區域二 且右二1側周緣91 ’該第二外側周緣91與圓環狀影像70 /、有-第—距離D2，該第二距離D2大於零即可，其尺 ϊ i 亦即，該_域9 Q之面積小於圓形環狀 二：面積’且該内侧區域9°不接觸該圓形環 、、列孔三圖至第五圖所示，藉由光學量測系統50量 之不同深度位置，域轉—深度位置之孔洞 1!丨，可得到ΐ數個圓環狀影像7〇(如第四、五圖所N-\ 下: The Laplacian algorithm is as follows: FM= H(1, . + Ζ") where, 4.+-沙-l'v -1) _ 4g· (·Υ - l'v) - g (·'. 乂V+Ρ -4g, (.v -1, ν -1)+20g·. (,ν - L v) -4^(. (Λ· -1, r+^ (·,·一一i) 一% (·γ_1· .r)-总(Λ·-υ'+ι) 201144749 The algorithm of Contrast is as follows: imn F\4 α Connasr min The above focusing calculus The method is a generally known method, and the detailed calculation contents thereof are not described in detail. The present invention is characterized in that a focus algorithm is used to calculate a peripheral region and an inner region of the captured annular image. The figure shows the manner in which the present invention defines the peripheral region of the annular image: it displays one of the annular image 70's which is a circular annular region of equal width (the fourth graphic oblique) a circle-shaped region having a H-side peripheral edge 81 m - a first-outer peripheral edge 82' of the first-inner peripheral edge 81 having a first distance di from the annular image 7〇, the first-inner peripheral edge 81 Having a wide yield W' between the first outer circumference 82 The field 80 covers the hole 61. Preferably, the circular ring image 70 is located between the first inner circumference 81 and the first outer circumference 82 (as shown in the fourth figure), but In addition, the image of the hole that is captured is an annular shape, and the peripheral region of the present invention is not limited to: an annular shape. In the fourth figure, the first inner circumference 81 is circular, and the first makeup is ,: Zhou ί 82 can be a circular, rectangular, polygonal or arbitrary geometric shape of the peripheral region defined by the present invention. The setting principle is d/, day: inner peripheral edge 81 and annular image 7Q - The first distance ruler + 廿 = the peripheral edge area 80 can cover the annular image 7 ,, and the shape size is not limited. The fifth figure in the area shows the inner side of the annular image defined by the present invention. The defined inner region 90 (the fifth illustrated oblique grid 201144749 line region) is located in the annular image 7, but the inner core 'but does not include the fine circumference, the inner region 2 and the right second 1 side periphery 91 ' The second outer circumference 91 and the annular image 70 /, have - the first distance D2, the second distance D2 The radius 即可 i is that the area of the _ field 9 Q is smaller than the circular ring shape 2: the area 'and the inner area 9° does not contact the circular ring, and the column holes 3 to 5 As shown, by the different depth positions of the optical measurement system 50, the hole 1~丨 of the domain-depth position can obtain a plurality of annular images 7〇 (such as the fourth and fifth figures)
不 用聚焦演算法分析每一圓環狀影像70之用绫F 域80,以得到複數個筮含瞀#^豕川之周緣區 位置㈣^ 數值及其對應的複數個第一 算數值之二：圖：示，該第一位置數值即為該第-演 值之U立置，同理，利用聚焦演算法分析 严 =70之内側區域9〇，以得到複 ： 其對應的複數個第二位置數值，如第七圖所示；= 置數值即為該第二演算數值之聚焦位 卜 第-演算數值代表孔洞61之上’“獲數個 置，該複触帛二料數賊纽 聚焦位置。 < 下孔（孔洞底部） 於聚===:::域:r聚焦演算_ 大聚焦演算數值位於聚焦位置圖側區域之最 位詈A 1Q1 β ,亦即，上孔聚焦 置為191. 8㈣’下孔聚焦位置為1〇 位置減去下孔聚焦位置，可得到-結果數值St 201144749 m-109. 3 /z m=82. 5 // m，因此，所量測之孔洞61之深度Η(如 第三圖所示）為82. 5//m。 _ 綜上所述，本發明提供之量測孔洞深度之方法，利用 跨焦影像的方式擷取孔洞不同深度位置的影像，並以聚焦 演算法（Focal Metric)分別分析上孔及下孔的聚焦位置， 可提高量測精確度，尤適用於量測高深寬比的孔洞深度。 惟以上所述者，僅為本發明之實施例而已，當不能以 之限定本發明所實施之範圍。即大凡依本發明申請專利範 .· 圍所作之均等變化與修飾，皆應仍屬於本發明專利涵蓋之 範圍内，謹請貴審查委員明鑑，並祈惠准，是所至禱。 201144749 【圖式簡單說明】 第一圖係習知聚焦演算數值與聚焦位置關係曲線圖。 第二圖係習知美國專利6882436號之系統架構圖。 第三圖係本發明之光學量測系統量測與待量測孔洞之 位置關係示意圖。 第四圖係本發明之周緣區域與環狀影像之關係示意 圖。 第五圖係本發明之内側區域與環狀影像之關係示意 圖。 第六圖係本發明之周緣區域聚焦位置曲線圖。 第七圖係本發明之内側區域聚焦位置曲線圖。 【主要元件符號說明】 先前技術： 10-待測工件 16-埋頭孔 20-影像偵測單元 26-照明單元 160-上孔距離 164-下孔距離 . 166-埋頭孔深度 本發明： 5〇-光學量測系統 12 201144749 60- 物件 61- 孔洞 70-圓環狀影像 80- 周緣區域 81- 第一内.側周緣 82- 第一外側周緣 90- 内側區域 91- 第二外側周緣 D1-第一距離 D2-第二距離 H-深度 W-寬度The 绫F domain 80 of each annular image 70 is analyzed without using a focus algorithm to obtain a plurality of 筮 瞀 豕 豕 豕 之 之 之 之 豕 豕 豕 豕 豕 豕 数值 数值 数值 数值 数值 数值 数值 数值 数值 数值 及其 及其 及其 及其 及其 及其 及其 及其 及其 及其 及其 及其 及其 : : : : indicates that the first position value is the U-stand of the first-value, and similarly, the inner region 9严 of the strict=70 is analyzed by the focus algorithm to obtain a complex: the corresponding plurality of second position values As shown in the seventh figure; = the value is the focus position of the second calculus value. The first-calculus value represents the upper part of the hole 61 '", and the number of points of the reticle is determined. < Lower hole (bottom of the hole) in the cluster ===::: field: r focus calculation _ The large focus calculation value is located at the most position 1A 1Q1 β of the side of the focus position map, that is, the upper hole focus is set to 191. 8(4) 'The lower hole focus position is 1〇 position minus the lower hole focus position, and the result value - St 201144749 m-109. 3 /zm=82. 5 // m, therefore, the depth of the measured hole 61Η (As shown in the third figure) is 82.5/m. _ In summary, the present invention provides a method for measuring the depth of a hole, using a cross The image of the focal image captures the image at different depths of the hole, and the focus position of the upper and lower holes is analyzed by Focal Metric, which can improve the measurement accuracy, especially for measuring the hole with high aspect ratio. The above is only the embodiment of the present invention, and the scope of the present invention cannot be limited thereto, that is, the equal variation and modification of the patent application according to the present invention should still be Within the scope of the patents covered by this invention, I would like to ask your review committee to give a clear understanding and pray for it. It is the prayer to be prayed. 201144749 [Simple description of the diagram] The first diagram is a plot of the relationship between the numerical values of the conventional focus and the position of the focus. The figure is a system architecture diagram of the prior art U.S. Patent No. 6,882,436. The third figure is a schematic diagram of the positional relationship between the optical measurement system measurement and the hole to be measured according to the present invention. The fourth figure is the peripheral region and the ring image of the present invention. Fig. 5 is a schematic diagram showing the relationship between the inner region of the present invention and the annular image. The sixth graph is a graph of the focus position of the peripheral region of the present invention. The inner region focus position curve of the present invention. [Main component symbol description] Prior art: 10-workpiece 16 to be tested - countersunk hole 20 - image detecting unit 26 - lighting unit 160 - upper hole distance 164 - lower hole distance. 166 - countersink depth The present invention: 5 〇 - optical measuring system 12 201144749 60 - object 61 - hole 70 - annular image 80 - peripheral region 81 - first inner side edge 82 - first outer circumference 90 - inner region 91- Second outer circumference D1-first distance D2-second distance H-depth W-width