200403716 玖、發明說明： 發明說明技術領域 本發明係關於利用半導體裝置之製程之微影工序施行 曝光用光罩與被曝光體之晶圓之相對位置之對準用之對 準裝置、對準方法及半導體裝置之製造方法。 先前技術 一般，在微影工序中所使用之電子線（Electr〇n Beam: EB)始接曝光裝置或χ線曝光裝置中，使用即時檢測曝光用 光罩與晶圓之相對位置而施行位置對準之對準裝置。此係 由於在曝光中也有可能發生曝光用光罩與晶圓之相對位 置之偏移，因此，需要即使地施行對準（位置對準）之故。 尤其在EB密接曝光裝置中，因需在真空中施行曝光，故會 發生因熱應變引起之位置漂移或裝置振動引起之位置晃 動等，致使發生上述相對位置之偏移之可能性升高之故。 發生位置漂移或位置晃動等之理由有後述①〜④項： ①由於減壓至真空，故會因隔熱膨脹導致裝置溫度降低。 〇因在真玄中曝光，故無法使用高精度之氣動滑閥。 ③在光微影工序中，雖可使用超殷鋼（超級因瓦鐵鎳合金） 等低膨脹合金，將裝置之熱膨脹應變抑制在較小值，但為 了穩定地保持電子線之方向，卻不能使用磁性體之低膨脹 合金。 ⑧在-真空中無空 而，在對準裝置中，最好使用不使晶圓上之光阻膜感j 之長波長光進行曝光。作為使用長波長光之對準光學 84634.doc 200403716 大致上可分為（1)使寫入之對準標記在曝光用光罩與晶圓 上成像之所謂成像方式、與（2)利用雷射干擾儀檢測寫入光 罩與晶圓之光柵產生之繞射波之相位之所謂干擾方式。 但，其中任何依種方式，為了避免對準光學系與曝光EB、 曝光X線或光罩載物台等機構零件相干擾，在其對準光學 系之光學零件之配置上都會受到限制。 特別在將在晶圓上以數十# m程度之間隔配置光罩，將 低能量之EB照射在光罩之技術之LEEpL(1〇w electr—on beam proximity projection lith〇graphy :低加速電 子線等倍近接曝光技術）適用於成像方式時，最好以數值孔 徑NA較大之物鏡，使其光軸垂直地接近於光罩與晶圓，以 便使對仏$己成像。但由於受到曝光裝置之空間限制，必 須使光軸傾斜40°之程度，NA也必須被限制於〇 35之程 度。另外，照明光學系也受到其空間限制，故難以由亮視 野照明或由可看到較亮之對準標記之暗亮視野照明之適 當方向進行照明。因此，例如特許第2955668號公報、特 許第3048904號公報及特許第3235782號公報中均提案使 用由觀察光學系照明之落射照明。 圖8係表示施行對準（位置對準）之際之對準標記與光與 系之配置之一具體例。如圖例所示，在施行對谁、 如以40°之入射角施行落射照明，以暗視野觀察對谁护、 5 1。而·，利用分別檢測曝光用光罩上及晶圓上之對谁把: 及晶圓之間隔之對準。又，在此所稱之入射面 51之位置，施行垂直於入射面之方向之對準血 ^ 曝先用光罩 ’係指物鏡 84634.doc 200403716 52之光軸與光罩•晶圓法線形成之面。 但，在上述（以往之對準中，為避免對準光學系與曝光 Μ或曝光X線等相干擾，必須使對_準標記51之照明光之 光軸大幅傾斜，因此，有發生以下之問題之虞。 例如以40。之入射角施行落射照明時，必須以肋。之大 角度，利iil 則方散射之光以暗視野觀察對準標記5工，其 結果，所得之像強度會變小。通常，對準標記51為矩形， 係由垂直於入射面之第1邊513與平行之第2邊5讣所構成。 若對準標記51充分小於波長，則不管來自第1邊51&之繞射 光或來自第2邊51b之繞射光也會各向同性地擴散。然而， 在達到與波長同程度以上之大小時，垂直於入射面之第丄 邊5 1 a之繞射光雖在入射面内約略各向同性地擴散，但平行 於入射面之第2邊5 lb之繞射光則呈現在與物鏡52相反侧 之正反射方向具有指向性。因此，入射於物鏡52之散射光 f使得第1邊5 la之繞射光大於第2邊5 lb之繞射光，結果第1 邊5 la之像強度會強於第2邊5 lb之像強度，因此，第1邊5 la 之像強度分布容易因標記角之形狀失真而變化，標記像之 強度分布也容易因標記角之形狀失真而變化，結果檢測誤 差有增大之虞。 對此’例如在上述各公報中都有利用配置多數第1邊5 1 a 之方式使其平均化之揭示。此係由於使第1邊5 1 a之位置在 長度方向平均化時，可高精度地測出標記位置之故。但， 配置多數第1邊5 1 a卻會招致對準標記5 1之複雜化及大型 化，使處理負荷也會因平均化等而增大，故並非好的作法。 84634.doc 200403716 、又、’例如將干擾方式適用於LEEPL時，連收敛角o irad 乂下之、，’田的雷射光束也能加以檢測，故可縮小對準光學 系，減少曝光裝置之零件與曝光卽相干擾。因此，比成像 方式更能縮小人射角，並增大所檢測之光量。但由於光拇 無聚焦之機構，在光罩與晶圓之間格方面有不能對準之 虞。、垂直於入射面之方向之位置呈現相當於光柵週期之整 數倍份之不確定性，故雖需要另外設置有別於本對準裝置 《粗《用之對準裝置’但卻可利用干擾儀檢測光拇之繞 射光<相位’以便高精度地施行對準。又，在晶圓上沉積 有光阻膜等時，有時，繞射光會因光柵之形狀而變小，此 時，有不能檢測繞射光之相位之虞。此點雖可利用使用多 數波長各異之雷射或使用波長可變之雷射之方式加以解 決，但會導致對準光學系之構造之複雜化及成本之提高， 故並非適當之作法。 另外，通常，LEEPL之光罩為補強其強度，都設置有格 子狀之樑，利用此樑將光罩分成多數小單元，因此，使光 軸大幅傾斜時，光束會被樑遮住，故可配置對準標記51之 區域也會侷限於不受該樑之影響之區域。具體而言，例如 如圖9所示，由於光軸之傾斜，會因垂直於該光軸之樑53& 之存在，使窝入區域被限制於朝向物鏡52之樑53b附近， 且因平行於光軸之樑53c而使窝入區域被限制於各小單元 之中央·附近。 因此，本發明之目的在於提供即使不使光軸大幅傾斜， 也不致於與曝光EB或曝光x線等相干擾，且可利用檢測對 84634.doc 200403716 準標圮又位置，咼精度地檢測其標記位置，結果可實現對 準之高精度化之對準裝置、對準方法。 發明内容 本發明係為達成上述目的所研發而成，其特徵在於··在 對準裝置中包含以光學方式檢測分別配設於曝光用光罩 及被曝光體之晶圓之對準標記之光學系，依據該光學系之 檢測結果，施行前述曝光用光罩與前述晶圓之相對位置之 對準，且在前述光學系之光路上包含縮小該光路之光束之 狹缝：與彎曲經過該狹縫後之光路之光軸方向之光軸變換 手段。 又，本發明為達成上述目的所研發之對準方法之特徵在 於·使用以光學方式檢測分別配設於曝光用光罩及被曝光 體之晶圓之對準標記之光學系，依據該光學系之檢測結 果，施行前述曝光用光罩與前述晶圓之相對位置之對準， 且利用在前述光學系之光路上設置之狹缝縮小該光路之 光束’利用前述狹縫彎曲光束被縮小後之光路之光軸方 向使4曲後之光路到達前述晶圓，藉以檢測前述對準標 記。 利用上述構成之對準裝置及上述步驟之對準方法，在例 如光學系使用ΝΑ=0·35程度之物鏡之情形，考慮其光束徑 時，僅依賴反射鏡及棱鏡，欲彎曲光軸雖有困難，但在該 情形下，也由於光學系之光路上之狹縫已經將該光路之光 束縮小，故利用反射鏡及棱鏡等光軸變換手段即可容易地 彎曲光束縮小後之光軸方向，因此，經由光軸之彎曲作 84634.doc -10- 200403716 用，即使不使對對準標記之入射角大幅傾斜，也可一面避 免與曝光EB及曝光X線或光罩載物台等機構零件等相干 擾，一面檢測對準標記之位置。且由於光束已經縮小，在 施行相對之位置對準之對準方向上，不會降低光學系之解 像度，故對準精度也不致於降低。 實施方式 以下，依據圖式說明本發明之對準裝置、對準方法及半 導體裝置之製造方法。但，以下說明之實施形態僅係實現 本發明之一例，本發明之内容當然不限定於此。 【第一實施形態】200403716 (1) Description of the invention: Technical description The present invention relates to an alignment device, an alignment method, and a method for aligning the relative positions of an exposure mask and a wafer to be exposed using a lithography process of a semiconductor device process. Manufacturing method of semiconductor device. In the prior art, in general, an electron beam (Electron Beam: EB) used in a lithography process is connected to an exposure device or a x-ray exposure device, and the relative position of the exposure mask and the wafer is detected in real time to perform position alignment. Align the device. This is because the relative position of the exposure mask and the wafer may shift during the exposure. Therefore, it is necessary to perform alignment (position alignment) even if it is performed. Especially in the EB close-contact exposure device, because exposure needs to be performed in a vacuum, position drift due to thermal strain or position shaking due to device vibration, etc., may increase the possibility of the above-mentioned relative position shift. . Reasons for position drift or position sloshing include the following items ① to ④: ① Because the pressure is reduced to a vacuum, the temperature of the device decreases due to thermal expansion. 〇Because of exposure in Shingen, high-precision pneumatic slide valves cannot be used. ③ In the photolithography process, although low-expansion alloys such as Super Invar (Super Invar) can be used to suppress the thermal expansion strain of the device to a small value, in order to stably maintain the direction of the electron beam, it cannot Low-expansion alloy using magnetic body. ⑧No air in -Vacuum In the alignment device, it is best to use long-wavelength light that does not make the photoresist film on the wafer sense. As alignment optics using long-wavelength light, 84634.doc 200403716 can be roughly divided into (1) a so-called imaging method in which a written alignment mark is imaged on an exposure mask and a wafer, and (2) a laser is used The so-called interference method in which the interference meter detects the phase of the diffraction wave generated by the grating written on the photomask and the wafer. However, in any of these ways, in order to avoid interference between the alignment optical system and the mechanical parts such as the exposure EB, the exposure X-ray, or the mask stage, the configuration of the optical components of the alignment optical system will be limited. In particular, the LEEpL (1〇w electr-on beam proximity projection lithography) technology, in which photomasks are arranged on the wafer at intervals of tens of #m, and low-energy EB is irradiated on the photomask, is a low-acceleration electron beam. When equal magnification is applied to the imaging method, it is best to use an objective lens with a large numerical aperture NA to make its optical axis perpendicular to the mask and the wafer in order to image the opposite lens. However, due to the space limitation of the exposure device, the optical axis must be tilted to an extent of 40 °, and the NA must also be limited to a degree of 0.35. In addition, the illumination optical system is also limited by its space, so it is difficult to illuminate the light from a bright field or from an appropriate direction of a dark light field where a brighter alignment mark can be seen. Therefore, for example, Japanese Patent No. 2955668, Japanese Patent No. 3048904, and Japanese Patent No. 3257782 have proposed the use of epi-illumination by the observation optical system. FIG. 8 shows a specific example of the arrangement of the alignment mark and the light system when the alignment (positional alignment) is performed. As shown in the figure, for whom, for example, the epi-illumination is performed at an incidence angle of 40 °, and the protection of the person is observed in a dark field. And, it is used to detect who is on the exposure mask and on the wafer, respectively: and the alignment of the wafer interval. In addition, at the position of the incident surface 51 referred to herein, alignment blood perpendicular to the direction of the incident surface is applied. ^ The first mask used for exposure refers to the optical axis of the objective lens 84634.doc 200403716 52 and the wafer normal. Formation face. However, in the above-mentioned (conventional alignment, in order to avoid interference between the alignment optical system and the exposure M or exposure X-ray, etc., the optical axis of the illuminating light of the collimation mark 51 must be greatly inclined. Therefore, the following occurs For example, when the epi-illumination is performed at an incident angle of 40 °, a rib must be used at a large angle. If the angle is high, the scattered light is observed in a dark field of view with the alignment mark. As a result, the intensity of the obtained image will change. Generally, the alignment mark 51 is rectangular and is composed of a first side 513 perpendicular to the incident surface and a parallel second side 5 讣. If the alignment mark 51 is sufficiently smaller than the wavelength, it does not matter whether it comes from the first side 51 & The diffracted light or the diffracted light from the second side 51b also diffuses isotropically. However, when it reaches a size equal to or greater than the wavelength, the diffracted light perpendicular to the first side 5 1 a of the incident surface is incident. The in-plane diffuses approximately isotropically, but the 5 lb diffracted light parallel to the second side of the incident surface shows directivity in the direction of regular reflection opposite to the objective lens 52. Therefore, the scattered light f incident on the objective lens 52 makes The diffracted light of 5 la on the first side is larger than 5 lb on the second side As a result, the intensity of the image on the first side 5 la will be stronger than the image intensity on the second side 5 lb. Therefore, the image intensity distribution on the first side 5 la is likely to change due to the shape distortion of the marker angle, and the intensity of the marker image The distribution is also likely to change due to the distortion of the shape of the mark angle, and as a result, the detection error may increase. In this regard, for example, there are disclosures in the above-mentioned publications in which a plurality of first sides 5 1 a are arranged and averaged. This is because when the position of the first side 5 1 a is averaged in the length direction, the mark position can be measured with high accuracy. However, arranging most of the first side 5 1 a complicates the alignment mark 51. The increase and increase in size cause the processing load to increase due to averaging, etc., so it is not a good practice. 84634.doc 200403716 ， 'For example, when the interference method is applied to LEEPL, even the convergence angle o irad 、, The laser beam of Tian can also be detected, so it can reduce the alignment optics, reduce the interference between the parts of the exposure device and the exposure frame. Therefore, it can reduce the angle of human exposure and increase the amount of light detected than the imaging method. . But because the thumb has no focus mechanism, There may be a misalignment between the mask and the wafer. The position perpendicular to the incident surface presents an uncertainty equivalent to an integer multiple of the grating period, so although it needs to be set separately from this alignment The device "coarse" is used for the alignment device, but the interference light can be used to detect the diffracted light < phase " of the light thumb in order to perform the alignment with high precision. In addition, when a photoresist film is deposited on the wafer, sometimes The diffraction light will become smaller due to the shape of the grating. At this time, the phase of the diffraction light may not be detected. Although this point can be solved by using lasers with different wavelengths or using lasers with variable wavelengths. , But it will cause the complexity of the structure of the alignment optics and increase the cost, so it is not appropriate. In addition, in general, LEEPL masks are provided with a lattice-shaped beam to enhance its strength. The beam is divided into many small units by this beam. Therefore, when the optical axis is greatly inclined, the beam is blocked by the beam. The area where the alignment mark 51 is arranged is also limited to the area not affected by the beam. Specifically, for example, as shown in FIG. 9, due to the tilt of the optical axis, the recessed area is limited to the vicinity of the beam 53 b facing the objective lens 52 due to the existence of the beam 53 & The beam 53c of the optical axis restricts the nesting area to the center and vicinity of each small unit. Therefore, an object of the present invention is to provide an optical axis that does not interfere with the exposure EB or exposure x-ray, etc., even if the optical axis is not greatly tilted, and can detect the position and position of the 84634.doc 200403716 standard and accurately detect it. Marking the position, as a result, an alignment device and an alignment method capable of achieving high-precision alignment. SUMMARY OF THE INVENTION The present invention has been developed to achieve the above-mentioned object, and is characterized in that: the alignment device includes an optical device for optically detecting an alignment mark of a wafer respectively disposed on an exposure mask and a subject to be exposed. According to the detection result of the optical system, the alignment of the relative position of the exposure mask and the wafer is performed, and the optical path of the optical system includes a slit that narrows the light beam of the optical path: and the bend passes through the narrow Optical axis conversion means of the optical axis direction of the optical path after sewing. In addition, the alignment method developed by the present invention to achieve the above-mentioned object is characterized in that an optical system is used which optically detects alignment marks of the wafers respectively disposed on the exposure mask and the exposed body, and according to the optical system As a result of the inspection, the relative positions of the exposure mask and the wafer are aligned, and the light beam on the optical path is reduced by using a slit provided on the optical path of the optical system. The direction of the optical axis of the optical path makes the optical path after the 4 curve reach the wafer, thereby detecting the alignment mark. By using the above-mentioned alignment device and the above-mentioned alignment method, for example, when an optical system uses an objective lens with NA = 0.35, when considering its beam diameter, it only depends on the reflector and prism. Difficult, but in this case, because the slit on the optical path of the optical system has already reduced the beam on the optical path, the direction of the optical axis after the beam reduction can be easily bent by using optical axis conversion means such as a mirror and a prism. Therefore, by bending the optical axis as 84634.doc -10- 200403716, even if the angle of incidence of the alignment mark is not greatly tilted, it is possible to avoid exposure to EB and exposure X-rays or other mechanical parts such as the mask stage. Equal phase interference, while detecting the position of the alignment mark. And because the light beam has been shrunk, in the alignment direction in which the relative position alignment is performed, the resolution of the optical system will not be reduced, so the alignment accuracy will not be reduced. Embodiments Hereinafter, an alignment device, an alignment method, and a method for manufacturing a semiconductor device according to the present invention will be described with reference to the drawings. However, the embodiment described below is only one example for realizing the present invention, and the content of the present invention is not limited to this. [First Embodiment]
兹舉LEEPL適用於成像方式之情形為例加以說明。首 先，說明本實施形態之對準裝置之概略構成，圖1A至圖1C 係本發明之對準裝置之第一實施形態之概略構成之一例 之模式圖。、 如圖1A所不，在此說明之對準裝置例如係在使用於微影 …序之EB在接曝光裝置（未予圖式）中，供作即時地檢測曝 光用光罩1與晶圓2之相對位置而施行位置對準之用。更詳 :之’係使窝人於曝光用光罩i及晶圓2之對準標記成像， 施行圖像處理而檢測標記位置，依據其檢測結果，使曝光 用光罩！與晶圓2中之一方或雙方向垂直於入射面之方向 移動（參照圖8)以施行此等之位置對準。 又，如圖1B所示，在i個對準裝置中，僅就垂直於入射 :之-方向施行位置對準，因此，假定細密接曝光裝置 ，搭載多數（例如4方向份）之對準裝置，而構成可利用此 84634.doc • 11 - 200403716 等對準裝置施行有關曝光用光罩1與晶圓2之間之χ-γ方 向、旋轉方向、倍率等之位置對準之狀態。 為施行此種位置對準，各對準裝置如圖1Α所示，作為以 光學方式檢測分別配置於曝光用光罩1及晶圓2用之光學 系，設置有光源11、視野縮小部12、聚光透鏡13、分束器 14、物鏡21、分束器 22、成像透鏡23a、23b、CCD(；Chai*ge Coupled Device :電荷耦合元件）攝影機24a、24b、狹缝3 1、 及稜鏡32。又，在光學系以外之部分，例如處理對準標記 之檢出結果之圖像處理系、及使曝光用光罩1與晶圓2中之 一方或雙方移動之調整承載台系，因與以往相同，故在此 省略在圖中之圖示及其詳細說明。 光源11、視野縮小部12、聚光透鏡13 '及分束器14構成 之照明光學系也大致與以往相同。作為光源丨丨，可使用利 用白色之氙燈、鹼性_化物燈等之柯勒照明。但因曝光裝 置系在真2中施行EB曝-光，故光源11最好構成配設於真空 容器外侧，例如利用光導器將照明光引導至該真空容器 内。 又，物鏡21、分束器22、成像透鏡23a、23b、CCD攝影 機24a、24b構成之觀察光學系也大致與以往相同。又，分 別設有2個成像透鏡23a、23b及CCD攝影機24a、24b係基於 以下所述之理由。對對準標記之光學系之光軸之傾斜小 時，曝光用光罩1上之對準標記（以下稱光罩標記）及晶圓2 上之對準標記（以下稱晶圓標記）之位置在像面上會分離， 欲以1個CCD攝影機同時檢測兩標記時，需要較寬之對準用 84634.doc -12- 200403716 面積，因此，為在較窄之面積同時檢測光罩標記與晶圓標 吾己’需利用分束器2 2將光分成2個光路，並準備2種成像透 鏡23 a、23b及CCD攝影機24a、24b，分別檢測光罩標記與 晶圓標記。因此’成像透鏡23a、23 b及CCD攝影機24a、24b 未必需要2種。 而，在本實施形態所述之對準裝置之大特徵在於光學系 具有狹缝31及棱鏡32之點上。 狹缝31係配置於分束器14及物鏡21之間，具有可使來自 照明光學系之照明光透過之矩形開口，利用該開口縮小來 自照明光學系之光路之光束徑。但，狹缝3丨係以其矩形開 口之長度方向沿著施行曝光用光罩丨與晶圓2之位置對準 4際 < 相對移動方向（即垂直於照明光學系之入射面之方 向）被配置。 棱鏡32係配置於物鏡21曝光用光罩丨之間，利用對經過 狹缝31後之祕之偏向.，使其光路之絲方向轉彎。即， 棱鏡32具有作為彎曲光路之光抽方向之光轴變換手段之 機能。但’作為光軸變換手段之機能未必需要使用棱鏡 32例如如圖2A至圖2C所示，使用多數反射鏡33a、33b 也無妨。 又此等棱鏡32或反射鏡33a、3补雖配置於曝光用光罩工 附近’但王要係以避開H暴光用光罩工之正上方之方式配 置’以戈妨礙對該曝光用光罩1之EB曝光（有時為X線曝光） 動作之進行。 利用以上構成《光學系所檢測之光罩標記及晶圓標記 84634.doc -13 - 200403716 係以個別對應於曝光裝置上之各對準裝置方式，事前配置 於曝光用光罩1上及晶圓2上，但例如如圖ic或圖2C所示， 至少與對準方向成直交之方向之大小L最好為又〜5；[(入 為照射光之波長）程度。更詳言之，照射光之波長為又，該 如射光之知、射角為<9時，大小L最好滿足L < α又/(2 Θ )(但 α 1 4)。此係由於此種大小容易使光返回物鏡2丨，並可極 力排除檢測時之光之指向性之影響，且在配置上無需特別 大之空間之故。 其次，說明有關利用以上所述之對準裝置所施行之對準 步騾，即本實施形態之對準方法。 在本貫施形態之對準方法中，如上所述，由於在光學系 之光路上設有狹缝31、棱鏡32或反射鏡33a、33b，故在施 行對準之際，首先，利用該狹缝3丨縮小光束徑，接著，利 用棱鏡32或反射鏡33a、33b彎曲光束，使彎曲後之光束到 達曝光用光罩1及晶圓而後，經由顯微鏡使光罩標記及 晶圓標兄在CCD攝影機24a、24b上成像，利用圖像處理求 出其位置’檢測曝光用光罩1與晶圓2間之偏移量。 此時，例如物鏡之ΝΑ=0·35程度時，僅將稜鏡32或反射 鏡33 a、33b置於該物鏡21與曝光用光罩1之間之情形，考慮 光束徑時’欲彎曲光軸有所困難。但，在本實施形態中， 由於將狹縫3 1置於顯微鏡之物鏡2 1之光瞳上，以縮小垂直 於入射-面之方向之光束徑，故可利用棱鏡32或反射鏡33a、 33b容易地彎曲光軸之方向。 因此，利用彎曲光軸之方向，可縮小對光罩標記及晶圓 84634.doc -14- 200403716 標記之傾斜角。即，利用彎曲光軸之方向，棱鏡32或反射 鏡33a、33b可在不妨礙EB曝光等之狀態下，使對光罩標記 及晶圓標記之光軸接近於垂直。而，在對光罩標記及晶圓 標記之光軸接近於垂直時，連帶地，來自該光罩標記及晶 圓標記之第2邊5 1 b(參照圖8)，即來自平行於入射面之邊之 繞射光之強度也會增強。 圖3 A至圖3B係表示點像強度分布之一具體例之說明 圖。同圖中’圖3 A係表示無狹缝3 1時之圓形光曈，圖3B 係表示有狹缝3 1時之矩形光曈之情形。在此，有狹缝3 1 時’使該狹缝3 1比光瞳徑充分地狹窄，而使光瞳形狀近似 於矩形。又，圖4A至圖4B係表示邊之像強度分布之一具體 例之說明圖。在圖例中，圖4A也表示圓形光瞳，圖4B也表 π矩形光瞳之情形。又，邊之長度為丨# m，顯微鏡之物鏡 表示ΝΑ=0·35，倍率1〇〇倍之情形。 如以上所述，依據本實施形態所述之對.準裝置及對準方 法，可利用在光路上設置狹缝3丨，使對光罩標記及晶圓標 記之光軸接近於垂直，故可增大該光罩標記及晶圓標記之 像鈿度，結果，可提高光罩標記及晶圓標記之位置檢測精 度。尤其，由於第2邊51b之對準方向之位置係利用在其長 度方向平均而求得，故可藉此平均化謀求位置檢測精度之 提南。 且此時之檢測結果之解像力，由於狹縫3丨之矩形開口之 長度万向沿著垂直於入射面之方向被配置，故即使在入射 面方向之解像力因縮小光束徑而降低，在垂直於入射面之 84634.doc -15 - 200403716 方向，即對準方向之解像力也不會降低。因此，即使可藉 狹缝31而彎曲光軸方向，也不會因該狹缝31而降低對準精 度。 又，因利用狹縫31縮小光束，並使對光罩標記及晶圓標 記之光軸接近於垂直，故即使如LEEPL<光罩般，曝光用 光罩1設有樑，光束被該樑遮住之情形也變少，可觀察較 寬之區域。因此，可確保較寬之可寫入光罩標記及晶圓標 記之位置，而且因光束被樑遮住之情形較少，故也可使用 NA較-大之物鏡21，同時也可期待提高解像力。 •【第二實施形態】 其次，說明本發明之第二實施形態。但，在此僅說明與 第一實施形態不同之點。圖5係本發明之對準裝置之第二 貝施形怨之概略構成之一例之模式圖。 在此就明之對準裝置異於第一實施形態之情形，顯微鏡 光軸垂直於曝光用光罩！及晶圓2。具體而言，如圖5所示， 作為祆測對準&圮之光學系，具有連接於光源之光導器 15卞直透鏡1 6、分束器1 7、物鏡2 1、成像透鏡23、CCD 攝影機24、壓電承物台25、狹缝31、反射鏡33&、33b。 在如此構成之對準裝置中，係利用以壓電承物台25移動 物鏡21之方式施行分別對光罩標記及晶圓標記之聚焦動 作。在該情形下，也可利用在光路上設置狹缝31，彎曲光 軸万向…以提高光罩標記及晶圓標記之位置檢測精度。此 點與第一實施形態相同。又，有關壓電承物台25之機能及 光學系t王要構成本身因與以往大致相同，故在此省略其 84634.doc -16- 200403716 詳細說明。 而，在本實施形態所述之對準裝置之大特徵在於狹缝31 之配置上。 如圖5所示，顯微鏡光軸垂直於曝光用光罩1及晶圓2 時’可用亮視野觀察光罩標記及晶圓標記，故例如光罩標 記或晶圓標記之至少一方為凹凸構造時，其觀察將變得非 常困難。即，對於凹凸構造之對準標記，施行如光軸傾斜 <情形般之暗視野觀察時，其對準標記之檢測較為容易。 因此，在本實施形態之對準裝置中，將狹缝3丨之矩形開 口配置成偏移光學系之顯微鏡光軸中心。更具體而言，係 將狹縫31之矩形開口配置於僅位於物鏡21之光瞳之單側 < 一半4位置。此時，矩形開口之端緣部分最好與物鏡2ι 之光瞳之中心一致（參照圖5)。 如此配置狹缝3 1時，即可用暗視野觀察光罩標記及晶圓 標記。即，利用使狹縫31之開口偏移光學系之顯微鏡光軸 中心，即使在曝光用光罩丨及晶圓2之入射角變成大致垂直 時，也可用暗視野觀察之凹凸構造之對準標記。因此，即 使對凹凸構造之對準標記之入射角變成大致垂直時，其像 強度也會增大而變得容易檢測，結果可提高光罩標記及晶 圓標記之位置檢測精度。也就是$，可利时視野觀察之 實現，避免對準精度之降低。 【策三實施形態】 其/人，3明本發明之第三實施形態。但，在此僅說明與 上述各實施形態不同之點。圖6係本發明之對準裝置之第 84634.doc 200403716 三實施形態之要部構成之一例之模式圖。 在此說明之對準裝置異於第一或第二實施形態之情 形’其大特徵在於使用半導體雷射（Laser Diode ; LD :雷 射一極體）等相干性之雷射光源11之點上。 邊又在光學系之光路上配置狹缝3 1時，開口會因該狹 、、’逢3 1而又窄，故會降低像之党度。為避免此亮度之降低， 有使用高亮度光源之必要。但，作為白色燈，並無亮度高 過一般廣泛被使用之氙（Xe)燈以上之光源。因此，在本實 施形態之對準裝置中，作為亮度高過白色燈之光源，使用 照射雷射光之光源，例如使用半導體雷射等相干性之光 使用相干性之光源11時，斑點雜訊有成為誤差原因之 虞。因此，此時有必要適當組合適用以下所述之①〜⑤中 之一項或多項之方式，除去斑點雜訊，以防止檢測誤差之 發生。即①將雷射光分割成多數光束，使該等光束具有波 列長度之光路長差，再將光束重疊。或②移動置於照明光 學系中之擴散板。或③使用相干長度較短之超亮度二極體 (Super Luminescent Diode : SLD)或④使用外部諧振器型ld 中振盈波i晉在1 〇nm以上之利特曼（Littman)雷射或利特露 (Littrow)(德國 Sacher Laser Technik GmbH公司製品）雷射 等。或⑤使用LD，將高頻白雜訊重疊驅動電源。 又’-使用相干性之光源11時，以空間的相干性較高之雷 射光照明時，為了成為相干成像，解像力也可能降低。為 避免此現象，例如如圖6所示，只要利用具有2片透鏡陣列 84634.doc -18- 200403716 18a又光學系使空間的相干性降低後加以照明即可。具體而 :，為除去波譜雜訊，以比CCD攝影機24之電場頻率充分 回又頻率，利用振動台1 8b使兩透鏡陣列1 8a振動，並以兩 振動台18b之振動不同步方式，使相位差保持7Γ /2程度以上 或成為不同頻率。振動振幅充分大於（fa/f〇bj)( λ /NAQbj)，且 充2小於透鏡陣列18a之元件透鏡徑。在此，f〇bj為物鏡21 ' 匕為構成透鏡陣列1 8a之元件透鏡之焦距，^為波 長’ NAGbj為物鏡21《να。 另外，為有效地使LD等之雷射光源u之射出光透過狹缝 31，/、要將枉面透鏡19置於該雷射光源丨丨之正後方，將光 束整形即可。 如以上所述，使用雷射光源丨丨作為光源燈時，即使開口 α物釦2 1 <光曈上之狹缝3丨而變窄，也可避免像之亮度降 ,、、’"果可增大像強度，使其檢測更為容易。也就是說， 可獒问光罩標圮及晶圓標記之位置檢測精·度，避免對準精 度之降低。又，利用降低照射薄膜光罩之光量，可防止薄 膜光罩之溫度上升’結果可縮小光罩之熱膨脹，提高曝光 圖案之描綠精度。 透過照明光與成像光之曝光用光罩1因在表背面之反射 光（干擾’使得其透光率會因波長而發生大變化。又，因 塗敷於晶圓2上之光阻膜等之關係，有時可能因特定波長 而使备圓標記之反射率接近㈣，使晶圓標記之像強度接 近於0 °為避免因此等原因造成無法觀察晶圓標記，作為 雷射光源11 ’最好準備數種波長不同之雷射(LD)，而選擇 84634.doc -19- 200403716 使用：圓標記之像強度最大之波長之雷射。 【第四實施形態】 二人/兄明本發明之第四實施形態。但，在此僅說明與 上述各實施形態不同之點。 主在此況月之對準裝置之大特徵在於：在第三實施形態之 丨目：、外作為雷射光源11，使用波長可變之色素雷射、 里貝石田射、者翠玉雷射等之點上。如此，欲適宜地選 擇晶圓標記：像強度最大之波長之雷射非常容易。 又-在本男她形悲之情形中，作為可實現之非相干化之 手&例如有①將提供波長長度以上之光路長差之光束重 ®或②使擴散板移動。或與第三實施形態之情形同樣地 ③使照明光學系之（陣列）透鏡振動之手段。 【第五實施形態】 其次，說明本發明之第五實施形態。但，在此僅說明與 上述各實施形態不同之點。 在此說明之對準裝置之大特徵在於：在第一或第二實施 形悲所过明之反射鏡33a、33b(參照圖2A至圖2C、圖5)係 七、斜角 了 受之 MEMS(Micro Electro Mechanical Systems: 微機電系統）傾斜反射鏡之點上。即，反射鏡33&、33b例如 如同可藉靜電力控制傾斜角之金屬薄膜反射鏡一般，除彎 曲光軸方向之機能外，並具有調整其彎曲角之偏向調整機 能。 - 如此’使用傾斜角可變之反射鏡33a、33|3時，可藉調整 该反射鏡3 3 a、3 3 b之傾斜角，任意移動在曝光用光罩}上及 84634.doc -20- 200403716 晶圓2上之觀察位置。更詳言之，傾斜角可變之方向為一 方向時’可使觀祭位置一維地移動；傾斜角可變之方向為 二方向時，可使觀察位置二維地移動。又，2個反射鏡33a、 3 3 b之傾斜角均可變時，即使各反射鏡為僅可向一方向調 1傾斜角之反射鏡，也可利用使互相之調整方向成直交， 以使觀察位置二維地移動。因此，若反射鏡33a、33b具有 偏向碉整機能，即可擴大光罩標記及晶圓標記之配置之自 由度’且可確保對此等之觀察之通用性及彈性等，故在提 高光罩標記及晶圓標記之位置檢測精度上，非常合適。 又，反射鏡33a、3 3b之偏向角較大而發生散焦現象時， 如苐一貝施形怨所述，只要利用壓電承物台2 5移動物鏡 2 1 (參照圖5)消除該散焦現象即可。 【第六實施形態】 其次，說明本發明之第六實施形態。但，在此也僅說明 與上述各實施形態不同之點。 在此說明之對準裝置之大特徵在於：在第一或第二實施 形怨所說明之反射鏡33a、33b(參照圖2A至圖2C、圖5)係 可凋整焦距之MEMS可變形反射鏡之點上。即，反射鏡 33a、33b例如如同在壓電薄片形成之雙壓電晶片上塗敷或 貼附反射鏡、或可藉靜電力控制曲率之金屬薄膜反射鏡等 一般，除彎曲光軸方向之機能外，並具有調整該光軸之焦 點位【之聚焦調整機能。 如此，使用可調整焦距之反射鏡33a、33b時，可藉調整 該焦點位置，以對應配置於不同位置之光罩標記及晶圓標 84634.doc -21- 200403716 記之需要。更詳言之，由於顯微鏡光軸呈傾斜狀態，故改 變焦點位置時，可改變觀察位置，結果可任意移動在曝光 用光罩1上及晶圓2上之觀察位置。因此，若反射鏡33a、33b 具有聚焦調整機能，即可擴大光罩標記及晶圓標記之配置 之自由度’且可確保對此等之觀察之通用性及彈性等，故 在提高光罩標記及晶圓標記之位置檢測精度上，非常合 適。 又，具有聚焦調整機能之MEMS可變形反射鏡雖也可設 置作為’曲工轴方向用之反射鏡3 3 a、3 3 b，但與此個別獨 互地’將其配置於例如物鏡21與（：€〇攝影機24a、24之間也 無妨。 【第七實施形態】 其次，說明本發明之第七實施形態。但，在此也僅說明 與上述各實施形態不同之點。 在此說明之對準裝置之大特徵在於··在第一或第二實施 形悲所忒明之反射鏡33a、33b(參照圖2A至圖2C、圖5)中， 其中一方係第五實施形態所說明之MEMS傾斜反射鏡，他 方係第六實施形態所說明之可變形反射鏡之點上。即，反 射鏡33a、33b除彎曲光轴方向之機能外，並具有調整其彎 曲角之偏向調整機能、與調整其光軸之焦點位置之聚焦調 整機能。 如此’組合使用傾斜反射鏡與可變形反射鏡時，即使在 七、斜反射鏡之偏向角較大而發生散焦現象時，也可利用可 又y反射鏡施行聚焦，適切地予以應付。即，任意移動在 84634.doc -22- 200403716 曝光用光罩1上及晶圓2上之觀察位置非常容易，因此，在 提高光罩標記及晶圓標記之位置檢測精度上，更為合適。 【第八實施形態】 其次，說明本發明之第八實施形態。但，在此僅說明與 上述各實施形態不同之點。圖7係本發明之對準裝置之第 八實施形態之概略構成之一例之模式圖。 在此說明之對準裝置如圖7所示，其大特徵在於··在第 一實施形態之情形以外，在物鏡21之光瞳之前設置電流反 射鏡14，利用該電流反射鏡34掃描對準標記之檢測位置之 點上。即，可利用電流反射鏡34實現偏向調整機能。此時， 利用中繼透鏡35a、3 5b使電流反射鏡34與物鏡21之光曈成 為共輛關係。 如此，即使使用電流反射鏡34時，也可利用該電流反射 鏡34施行掃描，使曝光用光罩1上及晶圓2上之觀察位置任 意移動。因此，在提高光罩標記及晶圓標記之位置檢測精 度上，非常合適。又，配設2組電流反射鏡34中繼透鏡35a、 3 5 b，使其各掃描方向成直交時，可使觀察位置二維地移 動，故在謀求位置檢測精度之提高上，更為合適。 又，在上述第五至第八實施形態中，為實現偏向調整機 能、聚焦調整機能，係以使用MEMS傾斜反射鏡、MEMS 可變形反射鏡、電流反射鏡等之情形為例加以說明，但如 其他技-術，例如將反射鏡配置於可擺動或移動之承物台上 而實現偏向調整機能及聚焦調整機能，當然也無妨。又， 在偏向調整機能、聚焦調整機能中，其光軸變換手段不僅 84634.doc -23 - 200403716 可利用反射鏡33a、33b構成，當然也可適用於利用棱鏡32 構成之情形。另外，利用多數反射鏡33a、33b構成光軸變 :奐手&時’“需要使其全部具有％向調整機&及聚焦調 正機把，/、要其中至少丨個具有偏向調整機能或聚焦調整 機也’即可移動上述之觀察位置。 圖10係表示本發明之半導體裝置之製造方法之流程。該 半導體裝置之製造方法係至少包含：使用以光學方式檢測 分別配設於曝光用光罩及被曝光體之晶圓之對準標記之 光學系，利用在前述光學系之光路上設置之狹縫縮小該光 路< 光束（S 101)，利用前述狹缝彎曲光束被縮小後之光路 之光軸方向（S1〇2)，使彎曲後之光路到達前述晶圓，藉以 檢測則逑對準標記（S 103)，依據前述光學系之檢測結果， 施行别述曝光用光罩舆前述晶圓之相對位置之位置對準 (S104)，經由前述曝光用光罩在前述被曝光體施行帶電粒 子線、或放射線、X線、極短紫外光、紫外光、或光線之 曝光（S105)之工序，並施行其後之半導體裝置之製造工序 (S106)。由於至少包含經由曝光用光罩在被曝光體施行電 子線、離子線等帶電粒子線、放射線、χ線、極短紫外光、 糸外光、或光線之曝光之工序，故適合於製造半導體裝 置，且可在不與電子線、或離子粒子線等曝光EB或曝光χ 線等相干擾之狀態下，檢測對準標記之位置。因此，本發 明在利用模板光罩等LEEPL之光罩施行曝光時，可望成^ 更有效之半導體裝置之製造手段。 如以上所說明，依據本發明之對準裝置及對準方法，由 84634.doc -24- 200403716 於利用狹缝縮小光學系之光束，故可容易地彎曲其光轴方 向’因此，經由光軸之彎曲，即使不使光軸大幅傾斜，也 可在不致於與曝光EB或曝光X線等相干擾之狀態下，檢剛 對準標記之位置，且因利用狹缝縮小光束，在對準方向 上’不會降低解像度。又，由於不必使光軸大幅頻斜，即 使在LEEPL之光罩等有格子狀樑存在時，也可極力抑制對 率標記之寫入區域受到限制，因此，依據本發明，可在不 會招致對準標記之複雜化之狀態下，高精度地檢測其標記 位置：結果即可實現對準之高精度化。 圖式簡單說明 圖1A至圖1C係本發明之對準裝置之第一實施形態之概 略構成之一例之模式圖，圖1A係光學系部分之概略構成之 側面圖，圖1B係其要部之平面圖，圖1C係表示對準標記之 平面圖。 、 圖2A至圖2C係本發明之對準裝置之第—實施形態之概 略構成之另-例之模式圖，圖2A係光學系部分之概略構成 之侧面圖，圖2B係其要部之平面圖，圖2(：係表示對準標記 之平面圖。 ' 圖3A至圖3B係表示點像強度分布之—具體例之說明 圖，圖3A係表示無狹缝之圓形光瞳之情形之_，㈣係表 示有狹缝之矩形光瞳之情形之圖。 圖4A至圖4B係表示邊像強度分布之—具體例之說明 圖，圖4A係表示無狹缝之圓形光瞳之情形之圖，圖4b係表 示有狹缝之矩形光曈之情形之圖。 84634.doc -25- 200403716 圖5係本發明之對準裝置之第二實施形態之概略構成之 一例之模式圖。 圖6係本發明之對準裝置之第三實施形態之要部構成之 一例之模式圖。 圖7係本發明之對準裝置之第八實施形態之概略構成之 一例之模式圖。 圖8係表示施行對準（位置對準）之際之對準標記與光學 系之配置之一具體例之模式圖（其一）。 圖9_係表示施行對準（位置對準）之際之對準標記與光學 系之配置之一具體例之模式圖（其二）。 圖1 0係表示本發明之半導體裝置之製造方法之流程。 圖式代表符號說明 1 光罩 2 晶圓 11 光源^ 12 視野縮小部 13 聚光透鏡 14.22 分束器 15 光導器 16 準直透鏡 17 分束器 18a 透鏡陣列 18b 振動台 19 柱面透鏡 84634.doc 200403716 21.52 物鏡 23a. 23b 成像透鏡 24a. 24b 攝影機 25 壓電承物台 31 狹缝 32 棱鏡 33a.33b 反射鏡 34 電流反射鏡 35a.35b 中繼透鏡 51 對準標記 51a 第1邊 51b 第2邊 53a〜53c 樑 84634.doc 27-The case where LEEPL is applicable to the imaging method is described as an example. First, a schematic configuration of an alignment device of this embodiment will be described. Figs. 1A to 1C are schematic diagrams of an example of a schematic configuration of a first embodiment of an alignment device of the present invention. As shown in Figure 1A, the alignment device described here is, for example, used in photolithography ... The EB is connected to the exposure device (not shown), and is used to detect the exposure mask 1 and wafer in real time. 2 relative position for position alignment. More details: ’is to image the nested person on the exposure mask i and the alignment mark of wafer 2 and perform image processing to detect the position of the mark. Based on the detection result, use the exposure mask! It is aligned with the position where the one or both directions of the wafer 2 are moved perpendicular to the incident surface (refer to FIG. 8) (see FIG. 8). As shown in FIG. 1B, among the i alignment devices, alignment is performed only in the direction perpendicular to the incidence:-direction. Therefore, it is assumed that the exposure device is closely connected, and a large number of alignment devices (for example, 4 directions) are installed. It is possible to use this alignment device such as 84634.doc • 11-200403716 to perform the alignment of the position of χ-γ direction, rotation direction, magnification, etc. between the exposure mask 1 and the wafer 2. To perform such positional alignment, each alignment device is shown in FIG. 1A. As an optical system for optically detecting the photomask 1 and the wafer 2 respectively, a light source 11, a field of view reduction section 12, Condensing lens 13, beam splitter 14, objective lens 21, beam splitter 22, imaging lenses 23a, 23b, CCD (; Chai * ge Coupled Device) cameras 24a, 24b, slit 31, and 稜鏡32. The parts other than the optical system include, for example, an image processing system that processes detection results of alignment marks, and an adjustment stage system that moves one or both of the exposure mask 1 and the wafer 2. It is the same, so the illustration and detailed description in the figure are omitted here. The illumination optical system composed of the light source 11, the field-of-view reduction section 12, the condenser lens 13 ', and the beam splitter 14 is also substantially the same as the conventional one. As a light source, Koehler illumination using a white xenon lamp, an alkaline lamp, or the like can be used. However, since the exposure device performs EB exposure-light in the true 2, the light source 11 is preferably configured to be arranged outside the vacuum container, for example, a light guide is used to guide the illumination light into the vacuum container. The observation optical system including the objective lens 21, the beam splitter 22, the imaging lenses 23a, 23b, and the CCD cameras 24a, 24b is also substantially the same as in the past. The reason for providing the two imaging lenses 23a and 23b and the CCD cameras 24a and 24b is as follows. When the tilt of the optical axis of the optical system of the alignment mark is small, the positions of the alignment mark (hereinafter referred to as the mask mark) on the exposure mask 1 and the alignment mark (hereinafter referred to as the wafer mark) on the wafer 2 are at The image surface will be separated. If you want to use a CCD camera to detect two marks at the same time, a wider alignment area is required. 84634.doc -12- 200403716 area, so in order to detect mask marks and wafer marks simultaneously in a narrow area I'm going to use the beam splitter 22 to split the light into two light paths, and prepare two types of imaging lenses 23 a and 23 b and CCD cameras 24 a and 24 b to detect the mask mark and wafer mark, respectively. Therefore, two types of imaging lenses 23a and 23b and CCD cameras 24a and 24b are not necessarily required. The alignment device described in this embodiment is characterized in that the optical system has a slit 31 and a prism 32. The slit 31 is arranged between the beam splitter 14 and the objective lens 21 and has a rectangular opening through which the illuminating light from the illuminating optical system can pass. The opening is used to narrow the beam path from the optical path of the illuminating optical system. However, the slit 3 丨 is aligned with the length of the rectangular opening along the exposure mask 丨 aligned with the position of the wafer 2 and the relative movement direction (that is, the direction perpendicular to the incident surface of the illumination optical system) Be configured. The prism 32 is arranged between the exposure mask 丨 of the objective lens 21, and uses the deflection of the secretion after passing through the slit 31 to turn the direction of the light path. That is, the prism 32 has a function as an optical axis conversion means for the light extraction direction of the curved optical path. However, the function of the optical axis conversion means does not necessarily require the use of a prism 32. For example, as shown in Figs. 2A to 2C, it is also possible to use a plurality of mirrors 33a and 33b. These prisms 32 or mirrors 33a and 3 are arranged near the masker for exposure, but Wang wants to arrange it so as to avoid directly above the masker for H exposure. EB exposure (X-ray exposure sometimes) of mask 1 is performed. Based on the above configuration, the mask marks and wafer marks detected by the optical system 84634.doc -13-200403716 are arranged on the exposure mask 1 and the wafer in advance in a manner corresponding to each alignment device on the exposure device. 2 but, for example, as shown in FIG. 2C or FIG. 2C, the size L at least in a direction orthogonal to the alignment direction is preferably ~ 5; [(in is the wavelength of the irradiation light) degree. To be more specific, the wavelength of the irradiated light is again. When the radiation angle is < 9, the size L preferably satisfies L < α and / (2 Θ) (but α 1 4). This is because the size is easy to return the light to the objective lens 2 丨, and the influence of the directivity of the light during detection can be eliminated as much as possible, and there is no need for a particularly large space in the configuration. Next, the alignment steps performed by the above-described alignment device, that is, the alignment method of this embodiment will be described. In the alignment method of the present embodiment, as described above, the slit 31, the prism 32, or the mirrors 33a and 33b are provided on the optical path of the optical system. Therefore, when performing the alignment, first, use the slit. The slit 3 丨 reduces the beam diameter, and then the prism 32 or the mirrors 33a and 33b are used to bend the beam, so that the bent beam reaches the exposure mask 1 and the wafer, and then the mask mark and the wafer target are placed on the CCD through a microscope. Images are formed on the cameras 24a and 24b, and the positions of the detection mask 1 and the wafer 2 are detected by image processing. At this time, for example, when the NA of the objective lens is about 0.35, only the 稜鏡 32 or the reflecting mirrors 33 a and 33 b are placed between the objective lens 21 and the exposure mask 1, and the beam diameter is taken into consideration when considering the beam diameter. The shaft is difficult. However, in this embodiment, since the slit 31 is placed on the pupil of the objective lens 21 of the microscope to reduce the beam diameter in the direction perpendicular to the incident-surface, the prism 32 or the mirrors 33a and 33b can be used. Easily bend the direction of the optical axis. Therefore, the angle of inclination of the mask mark and wafer 84634.doc -14- 200403716 mark can be reduced by using the direction of bending the optical axis. That is, by using the direction in which the optical axis is bent, the prism 32 or the mirrors 33a and 33b can make the optical axis of the mask mark and the wafer mark close to the vertical direction without hindering the EB exposure and the like. When the optical axis of the mask mark and the wafer mark is close to vertical, the second edge 5 1 b (see FIG. 8) from the mask mark and the wafer mark is taken together, that is, parallel to the incident surface. The intensity of the diffracted light on the sides will also increase. 3A to 3B are explanatory diagrams showing a specific example of a point image intensity distribution. In the same figure, FIG. 3A shows a circular beam when the slit 31 is not present, and FIG. 3B shows a rectangular beam when the slit 31 is present. Here, when there is a slit 31, the slit 31 is sufficiently narrower than the pupil diameter, so that the pupil shape is approximately rectangular. 4A to 4B are explanatory diagrams showing a specific example of the image intensity distribution of the edges. In the illustration, FIG. 4A also shows a circular pupil, and FIG. 4B also shows a case of a π rectangular pupil. In addition, the length of the side is # m, and the objective lens of the microscope indicates that NA = 0.35 and the magnification is 100 times. As described above, according to the alignment device and alignment method described in this embodiment, a slit 3 丨 can be used on the optical path to make the optical axis of the mask mark and wafer mark close to vertical, so it can be used. Increasing the image power of the mask mark and wafer mark, as a result, the position detection accuracy of the mask mark and wafer mark can be improved. In particular, since the position of the alignment direction of the second side 51b is obtained by averaging in the length direction, the averaging can be used to improve the position detection accuracy. And the resolution of the detection result at this time is because the length of the rectangular opening of the slit 3 丨 is arranged along the direction perpendicular to the incident surface, so even if the resolution in the direction of the incident surface is reduced by reducing the beam diameter, it is perpendicular to The 84634.doc -15-200403716 direction of the incident surface, that is, the resolution of the alignment direction will not decrease. Therefore, even if the optical axis direction can be bent by the slit 31, the alignment accuracy is not lowered by the slit 31. In addition, since the light beam is narrowed by the slit 31 and the optical axes of the mask mark and the wafer mark are made close to vertical, even if it is a LEEP < mask, the exposure mask 1 is provided with a beam, and the beam is blocked by the beam. There are fewer living conditions, and a wider area can be observed. Therefore, it is possible to ensure a wide position where the mask mark and the wafer mark can be written, and since the beam is rarely blocked by the beam, an objective lens 21 with a larger NA can also be used, and the resolution can be expected to improve. . [Second Embodiment] Next, a second embodiment of the present invention will be described. However, only differences from the first embodiment will be described here. Fig. 5 is a schematic diagram showing an example of a schematic configuration of a second Bescher-shaped grudge of the alignment device of the present invention. Here it is clear that the alignment device is different from the first embodiment, and the optical axis of the microscope is perpendicular to the exposure mask! And wafer 2. Specifically, as shown in FIG. 5, as the optical system of the speculative alignment & 具有, there are a light guide 15 connected to a light source, a straight lens 16, a beam splitter 1 7, an objective lens 2 1, an imaging lens 23, CCD camera 24, piezoelectric stage 25, slit 31, and mirrors 33 &, 33b. In the alignment device thus constituted, focusing operations on the mask mark and the wafer mark are performed by moving the objective lens 21 by the piezoelectric stage 25, respectively. In this case, it is also possible to use a slit 31 provided on the optical path to bend the optical axis universally to improve the position detection accuracy of the mask mark and wafer mark. This point is the same as the first embodiment. In addition, since the function of the piezoelectric stage 25 and the configuration of the optical system t are substantially the same as those in the past, detailed descriptions thereof are omitted here 84634.doc -16- 200403716. The alignment device described in this embodiment is characterized by the arrangement of the slits 31. As shown in FIG. 5, when the optical axis of the microscope is perpendicular to the exposure mask 1 and the wafer 2, the mask mark and the wafer mark can be observed with a bright field of view. Therefore, for example, when at least one of the mask mark or the wafer mark has an uneven structure. , Its observation will become very difficult. That is, it is easier to detect the alignment mark of the concave-convex structure when observing the dark field like the tilt of the optical axis < Therefore, in the alignment device of this embodiment, the rectangular opening of the slit 3 is arranged to be offset from the center of the optical axis of the optical microscope. More specifically, the rectangular opening of the slit 31 is arranged at one half of the pupil 4 of the objective lens 21. At this time, the edge portion of the rectangular opening should preferably coincide with the center of the pupil of the objective lens (see FIG. 5). When the slits 31 are arranged in this manner, the mask marks and wafer marks can be observed in a dark field. That is, by shifting the opening of the slit 31 from the center of the optical axis of the microscope, even when the incident angle of the exposure mask and the wafer 2 becomes approximately vertical, the alignment mark of the uneven structure can be observed in a dark field of view. . Therefore, even if the incident angle of the alignment mark of the uneven structure becomes approximately vertical, the image intensity will increase and it will be easy to detect. As a result, the position detection accuracy of the mask mark and the wafer mark can be improved. That is, $, which can be used to realize the visual field observation and avoid the reduction of the alignment accuracy. [Third Embodiment] The third embodiment of the present invention will be described. However, only the differences from the above embodiments will be described here. FIG. 6 is a schematic diagram showing an example of the configuration of the main part of the 84634.doc 200403716 third embodiment of the alignment device of the present invention. The alignment device described here differs from the first or second embodiment in that it is characterized by using a coherent laser light source 11 such as a semiconductor laser (Laser Diode; LD: Laser-Polar). . When the slit 31 is arranged on the optical path of the optical system, the opening will be narrowed due to the narrowness, ′, and 31. Therefore, the degree of image will be reduced. To avoid this reduction in brightness, it is necessary to use a high-brightness light source. However, as a white lamp, there is no light source with a brightness higher than that of a xenon (Xe) lamp that is generally widely used. Therefore, in the alignment device of this embodiment, as the light source with higher brightness than the white lamp, a light source radiating laser light is used. For example, when a coherent light source 11 such as a semiconductor laser is used and a coherent light source 11 is used, speckle noise may Become the cause of error. Therefore, at this time, it is necessary to appropriately combine and apply one or more of the following methods ① to ⑤ to remove speckle noise to prevent detection errors from occurring. That is, ① laser light is divided into a plurality of light beams so that these light beams have a difference in optical path length of the wave length, and then the light beams are overlapped. Or ② Move the diffuser placed in the Department of Lighting Optics. Or ③ Use a super luminescent diode (SLD) with a short coherence length, or ④ Use a Littman laser or laser with an external resonator type LD that is above 10 nm. Littrow (manufactured by Sacher Laser Technik GmbH, Germany), lasers, etc. Or ⑤ Use LD to superimpose high-frequency white noise on the driving power. Also, when the coherent light source 11 is used, when it is illuminated with laser light having high spatial coherence, the resolution may be lowered for coherent imaging. To avoid this, for example, as shown in FIG. 6, as long as the optical system is used to reduce the coherence of the space, it can be illuminated by using two lens arrays 84634.doc -18-200403716 18a. Specifically: In order to remove the spectral noise, the two lens arrays 18a are vibrated by using the vibration table 18b at a frequency sufficiently higher than the electric field frequency of the CCD camera 24, and the phases are made asynchronous by the vibration of the two vibration tables 18b. The difference remains above 7Γ / 2 or becomes a different frequency. The vibration amplitude is sufficiently larger than (fa / f0bj) (λ / NAQbj), and is smaller than the element lens diameter of the lens array 18a. Here, f0bj is the objective lens 21 ′, is the focal length of the element lenses constituting the lens array 18a, and ^ is the wavelength. NAGbj is the objective lens 21 <να. In addition, in order to effectively pass the light emitted from the laser light source u such as LD through the slit 31, it is only necessary to shape the beam 19 by placing the bevel lens 19 directly behind the laser light source. As described above, when using a laser light source as a light source lamp, even if the opening α object buckle 2 1 < the slit 3 on the light beam is narrowed, the brightness of the image can be prevented from decreasing. The result is increased image intensity, making detection easier. In other words, you can inquire about the position detection accuracy and accuracy of the mask mark and wafer mark to avoid a reduction in alignment accuracy. In addition, by reducing the amount of light irradiated to the thin film mask, the temperature of the thin film mask can be prevented from increasing. As a result, the thermal expansion of the mask can be reduced, and the accuracy of the green pattern of the exposure pattern can be improved. The mask 1 for exposure through the illumination light and the imaging light is greatly changed by the wavelength due to the reflected light (interference) on the front and back surfaces, and the photoresist film applied on the wafer 2 etc. In some cases, the reflectivity of the prepared round mark may be close to 因 due to a specific wavelength, and the image intensity of the wafer mark may be close to 0 °. In order to prevent the wafer mark from being unable to be observed due to other reasons, it is used as the laser light source. Ready to prepare several types of lasers with different wavelengths (LD), and choose 84634.doc -19- 200403716 to use: the laser with the wavelength of the image intensity of the circle mark is the largest. [Fourth Embodiment] Two people / brothers The fourth embodiment. However, only the differences from the above-mentioned embodiments will be described here. The main feature of the alignment device in this case is that in the third embodiment, the purpose is: as the laser light source 11 , Using a variable wavelength pigment laser, Ribe Ishida, Zuiyu laser, etc. In this way, it is easy to choose a wafer mark appropriately: the laser with the wavelength with the highest intensity is very easy. And-in this man In her sorrowful situation, Non-coherent hands & For example, ① the light beam providing the difference in optical path length above the wavelength length is heavy or ② the diffuser is moved. Or in the same way as in the third embodiment ③ the illumination optical system (array) Means for lens vibration. [Fifth Embodiment] Next, a fifth embodiment of the present invention will be described. However, only differences from the above-mentioned embodiments will be described here. The major feature of the alignment device described here is that: The first or second embodiment of the mirror 33a, 33b (refer to FIG. 2A to FIG. 2C, FIG. 5) which is obscure, is a MEMS (Micro Electro Mechanical Systems :) That is, the mirrors 33 &, 33b are, for example, metal thin-film mirrors that can control the inclination angle by electrostatic force, in addition to the function of bending the direction of the optical axis, and have the function of adjusting the deflection of the bending angle. When using the mirrors 33a, 33 | 3 with variable tilt angles, you can arbitrarily move them on the exposure mask} and 84634.doc -20- 200403716 by adjusting the tilt angles of the mirrors 3 3 a, 3 3 b. View on Circle 2 More specifically, when the direction of the variable tilt angle is one direction, the viewing position can be moved one-dimensionally; when the direction of the variable tilt angle is two directions, the observation position can be moved two-dimensionally. When the tilt angles of the two mirrors 33a and 3 3b are both variable, even if each mirror is a mirror that can only be adjusted by one tilt angle in one direction, it is possible to make the adjustment directions orthogonal to each other to make the observation position It moves in two dimensions. Therefore, if the mirrors 33a and 33b have a deflection function, the freedom of arrangement of the mask marks and wafer marks can be increased, and the versatility and flexibility of observation can be ensured. Therefore, it is very suitable for improving the position detection accuracy of mask marks and wafer marks. When the deflection angle of the mirrors 33a, 3 and 3b is large and the phenomenon of defocusing occurs, as described by Zhe Yibei, as long as the objective lens 2 1 is moved by the piezoelectric stage 2 5 (see FIG. 5), this is eliminated. Defocusing is sufficient. [Sixth Embodiment] Next, a sixth embodiment of the present invention will be described. However, only the differences from the above embodiments will be described here. The major feature of the alignment device described here is that the mirrors 33a, 33b (refer to FIGS. 2A to 2C, and FIG. 5) described in the first or second embodiment are MEMS deformable reflections with adjustable focus. On the mirror. That is, the reflection mirrors 33a and 33b are, for example, similar to the coating or attaching of a reflection mirror on a bimorph formed of a piezoelectric sheet, or a metal thin-film reflection mirror capable of controlling curvature by electrostatic force, etc., except for the function of bending the direction of the optical axis. , And has the function of adjusting the focus position of the optical axis [focus adjustment function. In this way, when using the focus-adjustable mirrors 33a and 33b, the focal position can be adjusted to correspond to the mask marks and wafer marks 84634.doc -21- 200403716 which are arranged at different positions. More specifically, since the optical axis of the microscope is inclined, when the focal position is changed, the observation position can be changed. As a result, the observation positions on the exposure mask 1 and the wafer 2 can be arbitrarily moved. Therefore, if the mirrors 33a and 33b have a focus adjustment function, the degree of freedom in the arrangement of the mask marks and the wafer marks can be expanded, and the versatility and flexibility of observation can be ensured. Therefore, the mask marks are improved. And the position detection accuracy of the wafer mark is very suitable. In addition, although a MEMS deformable mirror having a focus adjustment function may be provided as the mirrors 3 3 a and 3 3 b for use in the direction of the crank axis, they are individually and individually disposed on, for example, the objective lens 21 and (: Between the cameras 24a and 24. [Seventh Embodiment] Next, a seventh embodiment of the present invention will be described. However, only differences from the above-mentioned embodiments will be described here. A major feature of the alignment device is that one of the mirrors 33a and 33b (refer to FIGS. 2A to 2C and FIG. 5) which are clearly understood from the first or second embodiment, is the MEMS described in the fifth embodiment. The tilting mirror is the other point of the deformable mirror described in the sixth embodiment. That is, the mirrors 33a and 33b have the function of bending the direction of the optical axis and the deflection adjustment function to adjust the bending angle. The focus adjustment function of the focal position of its optical axis. In this way, when a tilt mirror and a deformable mirror are used in combination, even when the deflection angle of the seven or oblique mirrors is large and defocus occurs, it can be used. Mirror implementation Focus, cope with it appropriately. That is, it is easy to move the observation position on the exposure mask 1 and wafer 2 at 84634.doc -22- 200403716. Therefore, it is necessary to improve the position detection of the mask mark and wafer mark. [Eighth embodiment] Next, the eighth embodiment of the present invention will be described. However, only the differences from the above-mentioned embodiments will be described here. Fig. 7 is the first embodiment of the alignment device of the present invention. A schematic diagram of an example of a schematic configuration of the eighth embodiment. The alignment device described here is shown in FIG. 7 and has the major feature that, except for the first embodiment, a current reflection is provided in front of the pupil of the objective lens 21. Mirror 14, the current mirror 34 is used to scan the point of the detection position of the alignment mark. That is, the current mirror 34 can be used to achieve the bias adjustment function. At this time, the relay mirrors 35a and 35b are used to make the current mirror 34 and The light beam of the objective lens 21 becomes a car-to-vehicle relationship. In this way, even when the galvano mirror 34 is used, the galvano mirror 34 can be used for scanning, and the observation positions on the exposure mask 1 and the wafer 2 can be arbitrarily shifted. Therefore, it is very suitable for improving the position detection accuracy of the mask mark and the wafer mark. In addition, when two sets of current mirrors 34 relay lenses 35a and 3 5 b are provided so that their scanning directions are orthogonal to each other, The observation position is moved two-dimensionally, so it is more suitable for improving the position detection accuracy. In addition, in the fifth to eighth embodiments, in order to realize the bias adjustment function and the focus adjustment function, MEMS tilt is used. The case of mirrors, MEMS deformable mirrors, current mirrors, etc. will be described as an example, but if other technologies are used, for example, the mirrors are arranged on a swingable or movable stage to realize the deflection adjustment function and focus adjustment. Of course, the function can be anyway. In addition, in the bias adjustment function and the focus adjustment function, the optical axis conversion means is not only 84634.doc -23-200403716, which can be configured by using the mirrors 33a, 33b, but of course, it can also be applied to the configuration using the prism 32. situation. In addition, the majority of the mirrors 33a and 33b are used to form the optical axis change: when the hand & "" needs to make all of them have a% adjustment machine & focus adjustment handles, /, at least one of them must have a bias adjustment function It is also possible to move the above-mentioned observation position with a focus adjustment machine. Fig. 10 shows the flow of a method for manufacturing a semiconductor device according to the present invention. The method for manufacturing the semiconductor device includes at least: using optical detection to be separately arranged for exposure The optical system of the alignment marks of the mask and the wafer of the exposed object uses the slit provided on the optical path of the optical system to narrow the optical path < the light beam (S 101), and uses the slit to bend the light beam to be reduced. The optical axis direction of the optical path (S102) enables the curved optical path to reach the aforementioned wafer, thereby detecting the alignment mark (S103), and performing another type of exposure mask according to the detection results of the aforementioned optical system. The relative positions of the wafers are aligned (S104), and charged particle rays, radiation, X-rays, extremely short ultraviolet light, ultraviolet light, or light rays are applied to the exposed object through the exposure mask. The process of exposure (S105), and the subsequent manufacturing process (S106) of the semiconductor device. Since it includes at least the application of charged particle rays, radiation, x-rays, etc. to the object to be exposed through the exposure mask. Ultra-short ultraviolet light, external light, or light exposure process, so it is suitable for manufacturing semiconductor devices, and can be detected without interfering with electron beams, ion particle beams, and other exposure EB or exposure x-rays. The position of the alignment mark. Therefore, the present invention is expected to be a more effective method for manufacturing semiconductor devices when performing exposure using a mask of LEEPL, such as a stencil mask. As explained above, the alignment device and The alignment method is based on 84634.doc -24- 200403716. The optical beam is narrowed by using a slit, so the optical axis can be easily bent. Therefore, even if the optical axis is not tilted significantly, the optical axis can be bent through the bending of the optical axis. In a state where it does not interfere with the exposure EB or exposure X-ray, etc., the position of the just-aligned mark is detected, and the beam is narrowed by using a slit, and the resolution is not reduced in the alignment direction. Since it is not necessary to make the optical axis significantly inclined, it is possible to minimize the restriction of the writing area of the contrast mark even in the presence of a lattice beam such as a mask of LEEPL. Therefore, according to the present invention, alignment can be avoided without incurring When the mark is complicated, the mark position can be detected with high accuracy: As a result, the high precision of the alignment can be achieved. The drawing briefly illustrates that FIG. 1A to FIG. 1C are outlines of the first embodiment of the alignment device of the present invention. Fig. 1A is a schematic view of an example of the constitution, Fig. 1A is a side view of a schematic constitution of an optical system part, Fig. 1B is a plan view of a main part, and Fig. 1C is a plan view showing an alignment mark. Figs. 2A to 2C are a pair of the present invention. Fig. 2A is a side view of the schematic structure of the optical system part, Fig. 2B is a plan view of the main part, and Fig. 2 (: shows the alignment mark Floor plan. 'Figures 3A to 3B show the intensity distribution of a point image—an explanatory diagram of a specific example, and FIG. 3A shows a case of a circular pupil without a slit, and FIG. 3A shows a case of a rectangular pupil with a slit. Illustration. 4A to 4B are explanatory diagrams showing the intensity distribution of the side image—a specific example, FIG. 4A is a diagram showing a case of a circular pupil without a slit, and FIG. 4b is a diagram showing a case of a rectangular pupil having a slit Illustration. 84634.doc -25- 200403716 Fig. 5 is a schematic diagram showing an example of a schematic configuration of a second embodiment of the alignment device of the present invention. Fig. 6 is a schematic diagram showing an example of a configuration of a main part of a third embodiment of the alignment device of the present invention. Fig. 7 is a schematic diagram showing an example of a schematic configuration of an eighth embodiment of the alignment device of the present invention. FIG. 8 is a schematic diagram (part 1) showing a specific example of the arrangement of the alignment mark and the optical system when the alignment (position alignment) is performed. Fig. 9_ is a schematic diagram (part 2) showing a specific example of the arrangement of the alignment mark and the optical system when the alignment (position alignment) is performed. FIG. 10 is a flowchart showing a method of manufacturing a semiconductor device according to the present invention. Explanation of Symbols in the Drawings 1 Mask 2 Wafer 11 Light Source ^ 12 Field of View Reduction 13 Condenser 14.22 Beam Splitter 15 Light Guide 16 Collimator Lens 17 Beam Splitter 18a Lens Array 18b Vibration Table 19 Cylindrical Lens 84634.doc 200403716 21.52 Objective lens 23a. 23b Imaging lens 24a. 24b Camera 25 Piezo stage 31 Slit 32 Prism 33a.33b Reflector 34 Current mirror 35a.35b Relay lens 51 Alignment mark 51a First side 51b Second side 53a ~ 53c Beam 84634.doc 27-