EXPOSURE CONTROL IN PHOTOLITHOGRAPHIC DIRECT EXPOSURE METHODS FOR PRINTED BOARD OR CIRCUIT PRODUCTION

Description

[0001] The invention relates to a device for exposure control in photolithographic direct exposure processes for two-dimensional structures in photosensitive coatings, preferably on printed circuit boards, display substrates or wafers, as well as a method for converting registration data into direct exposure data, in particular for adapting the registration to uneven substrates and their implementation as a result of a registration "on the fly".

[0002] "On-the-fly" registration here means that the position data of target marks and thus the position of the substrate to be processed and its unevenness are successively recorded directly during the continuous movement of the substrate and are made available for the alignment of the exposure pattern to the substrate by adjusting the exposure data for the immediately following direct exposure.

[0003] The field of application of the invention lies particularly in the electronics industry and in the semiconductor industry in the production of printed circuit boards, displays and chips.

[0004] Exposure systems for disc- or plate-shaped workpieces are known from the prior art. These systems can expose an exposure object to a predetermined pattern using electromagnetic radiation, predominantly in the visible or ultraviolet spectral range, a laser beam, or an electron beam or particle beam. Exposure only begins after the correct positional relationship has been established between the exposure object and its markings (targets) and a predetermined pattern stored in the exposure device. For this purpose, the targets on the exposure object are captured by one or more cameras, and the exposure object and exposure pattern are aligned with each other before or during the exposure.

[0005] For the production of conductor tracks or extremely small electronic structures on plate-shaped substrates, such as circuit boards, display substrates, or wafers, the exposure processes that must be carried out with high spatial precision and the necessary handling and alignment times of the plate-shaped workpieces are the limiting factors for increasing throughput. Therefore, the aim is to perform the handling and exposure steps overlappingly or simultaneously and, if the front and back sides are to be exposed in the same device, to shorten the downtime of the exposure process. Such solutions are disclosed, for example, in EP 0 951 054 A1, EP 0 722 123 B1, US Pat. No. 6,806,945 B2, and JP 2010-181519 A.

[0006] A further challenge with direct exposure methods is to enable registration independent of the type, number and position of the targets on the exposure object as well as of height deviations of the object surface.

[0007] A solution for capturing surface topographies is known from WO 2016 / 115536 ​​A2. In this method, a known two-dimensional pattern, projected onto the surface or otherwise applied, is recorded two-dimensionally together with the surface, and the three-dimensional shape of the surface is determined based on the distortion of the pattern caused by surface irregularities. However, due to its limited resolution, this method is unsuitable for capturing poorly differentiated irregularities.

[0008] EP 0 954 768 B1 describes a device for focusing on the surfaces of a semiconductor wafer. The device records the actual surface topography of a semiconductor wafer prior to exposure in an exposure system. For this purpose, the surface is recorded using a distance sensor to obtain height information, and periodic height fluctuations are then determined and stored. Based on the determined periodic height fluctuations, an optimized focus position for the exposure optics is determined for the partial areas of the wafer surface to be exposed, and the wafer is aligned accordingly. A disadvantage is the time-consuming acquisition of the height information using the distance sensor and the subsequent alignment of the wafer.

[0009] WO 03 / 094582 A2 discloses another registration control system for exposing substrates, particularly for laser direct exposure on multilayer PCBs (printed circuit boards). A digital control image is generated by non-uniformly modifying a representation of the electrical circuit such that an electrical circuit pattern imprinted on a substrate using the digital control image precisely matches an existing circuit part. For this purpose, selected reference marks are registered on an existing actual structure. Based on deviations between actual and target positions in the spatial directions, the target structure to be exposed is corrected, so that exposure occurs with a modified scanning grid. Since a camera integrated into the exposure head is used for registration, each reference mark must be approached by relative movement between the substrate and the scanning head, which is a disadvantage.

[0010] DE 10 2018 132 001 A1 discloses a device for processing plate-shaped workpieces with a high workpiece throughput for use in the direct exposure of printed circuit boards. In this device, the registration unit is equipped with two to three area scan cameras that can be moved laterally to the circuit board movement in order to detect target marks when the position of the target marks on the circuit board is known in advance. The cameras are arranged parallel to a processing path, alternating between two tables moving on the same rail system, in parallel to a processing path, in order to minimize cycle times for processing the printed circuit boards by shortening handling and non-productive times. The cameras are regularly arranged above the edge areas of the printed circuit boards or circuits of a printed circuit board panel in which the target marks are known to be located.The detection of arbitrarily positioned target marks, which are increasingly necessary for printed circuit boards and wafers, is only possible with reduced throughput due to the necessary camera shifts, and height fluctuations of the substrate surface cannot be detected at all.

[0011] EP 2 775 349 A1 describes a method for determining a correct focus position in an optical inspection system. A difference between the focus position of the inspection system and the position of an object to be inspected is determined, whereby an image of the object is captured regardless of the correct focus position. An algorithm of the inspection system can use features of the captured image to determine the magnitude of the difference and direction of the deviation between the focus position and the position of the object by adapting the position of the object to the focus position of the inspection system according to the difference and direction.In the procedure described for a sample to be examined under a microscope, in which various characteristic objects with different spatial dimensions and thus significant height differences within the observed area are to be located, a singular adjustment of the focus position is required to clearly capture the various objects. Capturing the height profile across the entire area is not planned.

[0012] The patents US 6,245,585 B1 and US 6,449,029 B1 describe methods and devices for adjusting the focus position during photolithography of a semiconductor wafer. Before exposure, the height of the surface in the z-direction is measured in each individual subsection of the wafer to be exposed. The subsections are rectangles arranged in a grid-like manner, which are exposed one after the other. The reflections of five obliquely incident laser beams, one of which is directed at the center and four of which are directed at each of the corners of the subsection, are recorded on the surface of each subsection. From the position of the reflections, offset values ​​by which the height of each subsection deviates from a reference height can be determined, and a surface angle can be calculated for each subsection.Before each section is exposed, the wafer is aligned according to the stored values. Alignment is performed using individually controllable actuators that adjust the height and angle. However, only an averaged correction across the entire section can be set for each section.

[0013] The unpublished DE 10 2019 128 198.9 describes a device for patterning a wound continuous substrate using radiation, in which the target registration and pattern exposure take place on a continuous substrate guided tightly on a processing drum. If the registration unit and processing unit are located diametrically opposite each other on the drum, a change in the height of the substrate due to the drum curvature can be used to adjust both the focusing of the cameras of the registration unit and the processing beam by moving the drum longitudinally to the direction of substrate movement. However, measuring the extent of the defocusing requires additional tools with which the absolute position of the substrate surface can be determined.

[0014] US 2004 / 0223129 A1 discloses an exposure device for the surface exposure of photosensitive material, which is moved in a plane relative to an exposure device comprising a plurality of identical exposure heads arranged in a matrix. In each of the exposure heads, the light from a light source is modulated into a two-dimensional pattern, which is projected onto the material surface for exposure using a telecentric lens. To compensate for unevenness at various locations on the material surface, each exposure head in the beam path has a pair of wedge prisms with which the optical length between the lens and the material surface is adjusted when distance sensors based on reflected laser radiation have detected unevenness due to a change in the location of the material surface.The disadvantage is the individual tracking for each exposure head and its image field as a single image segment, which can lead to sudden changes in the image scale in neighboring image segments.

[0015] The invention is based on the object of finding a new method for improved exposure control in direct exposure processes for two-dimensional structures in photosensitive layers on circuit boards or wafers, which allows for on-the-fly registration of targets independent of defined target locations and avoids the use of expensive telecentric lenses. An extended object is to achieve flexible adaptation of the exposure pattern even for detected substrate irregularities.

[0016] DE 10 2017 102 320 A1 discloses a processing system, in particular an optical processing system, for a substrate body, comprising an exposure system with one exposure unit or a plurality of exposure units, a calibration system with at least one calibration camera for adjusting the exposure system, a substrate carrier unit with a holding device for the substrate body and a registration system with one registration camera or a plurality of registration cameras, wherein a position and / or orientation of the substrate body held by the holding device in at least one registration position of the substrate carrier unit can be detected by the one or at least one registration camera, to be designed such that the exposure system and the registration system can be easily coordinated with one another, it is proposed that the calibration system has one or a plurality of reference markings,which are each arranged in a defined relative position to the at least one calibration camera, and that the one or at least one reference marking can be detected by the one or at least one registration camera.

[0017] US 2007 / 242317 A1 discloses an exposure system similar to DE 10 2017 102 320 A1.

[0018] US 2012 / 327215 A1 discloses an optical inspection sensor. The sensor includes a camera assembly configured to capture image data relative to a workpiece that is continuously moving relative to the camera assembly. An illumination system is arranged to deliver an illumination pulse when the camera assembly captures the image data.

[0019] According to the invention, the object is achieved in a device for exposure control during photolithographic direct exposure of two-dimensional structures in photosensitive coatings on a substrate, comprising a registration unit for registering target marks located on a substrate surface, a movable table system for supporting and defined one-dimensional movement of the substrate below the registration unit, a processing unit with a controllable linear processing path for photolithographic processing of the substrate by means of a processing beam for imprinting the two-dimensional structures, and a computer unit for controlling the alignment between the processing path and the substrate by means of local adaptation of the photolithographic processing depending on the position of the substrate determined by registered target marks,that in the registration unit, several entocentric cameras are arranged in a linear alignment transverse to the one-dimensional movement of the substrate to form a gapless linear scanning area over a predetermined width of the substrate and have extended image angles in the direction of the linear scanning area, wherein the image angles of adjacent entocentric,

[0020] Cameras along the linear scanning area have an overlap area in order to capture redundant image recordings of the substrate of the adjacent cameras in the overlap area, and in that the computer unit has means for calculating the position of the target marks from the redundant image recordings in the overlap area (13) of the adjacent entocentric cameras with additional use of a height position of the target marks determined by triangulation of a distance of the substrate surface.

[0021] Advantageously, the registration unit is equipped with a plurality of entocentric cameras for generating the linearly continuous, gapless sensor area such that image angles of adjacent cameras have an overlap area that is at least as large as half the image angle, wherein the computer unit is set up to determine target marks positioned arbitrarily across the width of the substrate, independently of the position of the target mark, within continuously successive overlap areas of the image angles of adjacent entocentric cameras by triangulation of a distance at any position on the substrate surface.

[0022] The cameras are preferably line scan cameras in order to make the linear scanning area narrow, seamless and with overlapping areas across the entire width of the substrate using cameras with a long scanning length.

[0023] It is advisable to point the cameras at the substrate surface with their optical axes parallel to each other, with the overlapping areas of the image angles of all cameras being the same.

[0024] In another advantageous embodiment, two adjacent cameras with mutually inclined optical axes are directed onto the substrate surface, wherein the overlapping area of ​​the image angles of the mutually inclined cameras is set such that the image angles of the two cameras completely overlap on the substrate surface.

[0025] The overlapping area, which is formed by pairs of cameras inclined towards each other, is connected without gaps to at least one further overlapping area until the overlapping areas have an extent that corresponds at least to the width of the substrate, wherein an overlap can be provided between pairs of cameras inclined towards each other in order to guarantee a gap-free scanning area of ​​the registration unit for all permissible height fluctuations Δz of the substrate surface.

[0026] Preferably, the cameras are arranged in pairs with each other tilted so that they are subject to a Scheimpflug condition.

[0027] In an advantageous embodiment of the invention, the computer unit additionally has a control for rapid focus tracking of the processing unit along the processing path as a function of height fluctuations Δz of the substrate surface, which includes a triangulation of target marks or any imaged structures of the substrate surface on the basis of the redundant image recordings in the overlapping area of ​​the adjacent entocentric cameras.

[0028] In a further preferred variant, the registration unit for target mark detection has only two entocentric cameras arranged on a scanning line (transverse to the direction of movement of the substrate) with an overlap range of one hundredth to one third of the image angle of the camera when the substrate is guided as a flexible endless substrate on a roller table system taut and without height fluctuations Δz, wherein the overlap range of the image angles of the two cameras is designed such that the triangulation for the purpose of precisely determining the thickness of the substrate in the overlap range of the image angles on the roller table system can be used and can be assumed to be constant for the entire width of the roller table system.

[0029] Furthermore, the registration unit is conveniently equipped with light sources for illuminating the linear, gapless scanning area, which are evenly distributed in the housing to create a scanning line homogeneously illuminated by dark field or bright field illumination.

[0030] The light sources are designed for continuous illumination and have a device for controlling at least one property of brightness, angle of incidence or spectral range.

[0031] In addition, the light sources can be set up for continuous illumination to enable image capture by controlling the integration time of the sensor line using an electronic shutter principle.

[0032] In a further preferred embodiment of the invention, a focus tracking system for height fluctuations Δz of the substrate is integrated into the processing unit, with which height fluctuations Δz determined by the computer unit by means of triangulation from images redundantly recorded with two adjacent cameras of the registration unit can be adapted in real time by rapid focus correction for each pixel of the registration unit, wherein the focus tracking can be controlled on the basis of changes in a lens or mirror position or curvature.

[0033] In this case, the focus tracking is advantageously controllable based on the change in a mirror curvature at least in the transverse direction x to the direction of movement y of the substrate. The focus tracking can preferably be controlled separately based on changes in a mirror curvature in the direction of movement y of the substrate and a mirror curvature in the transverse direction x.

[0034] In a practical design, the focus tracking can be controlled by changing the mirror curvature using a piezo element.

[0035] Advantageously, focus tracking based on changes in a lens or mirror position or curvature can also be used to correct imaging-related focus deviations of an upstream focusing optic or other upstream optical elements resulting from the optics design or optics manufacturing.

[0036] The object is further achieved by a method for exposure control in photolithographic direct exposure of two-dimensional structures in photosensitive coatings on a substrate, comprising the following steps: Arranging a plurality of entocentric cameras to form a continuous linear scanning area transverse to a direction of movement of the substrate in a registration unit for detecting target marks located on the substrate, wherein the entocentric cameras have extended image angles along the linear scanning area with an overlap area formed by adjacent entocentric cameras in order to obtain redundant image recordings of the substrate from adjacent cameras in the overlap area, Moving the substrate on a movable table system in a defined one-dimensional movement below the registration unit, Providing a processing unit for photolithographically generating the two-dimensional structures with a processing beam that is controllable along a linear processing path, Detecting the spatial position of target marks distributed arbitrarily over a predetermined width of the substrate with respect to length,Width and height positions during a single pass of the substrate through the linear scanning range of the registration unit, determining the positions of the target marks distributed arbitrarily across the width of the substrate from the redundant image recordings in the overlapping area of ​​the adjacent entocentric cameras, additionally using a height position of the target marks determined by triangulating a distance of the substrate surface from the redundant image recordings of adjacent entocentric cameras, calculating data for the alignment and local adjustment of the processing of the substrate with two-dimensional structures for the processing unit to control the processing beam along the linear processing path aligned transversely to the direction of movement of the substrate,and controlling the alignment between the processing path and the substrate and locally adapting the photolithographic processing depending on the position of the substrate determined by registered target marks. ,

[0037] In a preferred method variant, the calculation of the spatial position of target marks distributed arbitrarily across the width of the substrate on the basis of a height position during a pass through the substrate is extended to the triangulation of further detectable structures of the substrate in the redundant images recorded in the overlap area and a rapid focus adjustment of the focus of the processing beam takes place by means of focus tracking along a processing path on the basis of a control of a lens or mirror position or a mirror curvature.

[0038] The rapid focus adjustment of the focus of the processing beam is advantageously carried out at a frequency that is at least two to thirty times higher than the conventional scanning frequency for the processing beam.

[0039] The invention is based on the fundamental idea that one or more cameras with two-dimensional sensors (e.g. CCD cameras, CMOS cameras) are used for so-called registration systems. These cameras are usually equipped with telecentric lenses and arranged exactly vertically over defined sections of substrates. As a result, the detected positions remain constant within the available depth of field of telecentric cameras, even if the focus position of the lens relative to the substrate shifts due to changes in thickness or topography. Due to their design, telecentric lenses are comparatively expensive and so bulky that the mechanical dimensions of a lens must always be larger than the image field to be recorded. Telecentric lenses therefore do not allow continuous image recordings by several cameras positioned along a straight line, but would have to be arranged offset along several parallel lines.In addition, an inherent determination of height deviations of the substrate, which is essential for the highly accurate determination of the target position as well as for the precise alignment and adjustment of the exposure pattern, is not possible.

[0040] The invention solves these problems by a combination of a type of line-shaped image scanning across the entire substrate width (hereinafter: FPSS - English: F ull P ring S can System) with a progressive relative movement between the substrate and the linear scanning area by cameras with entocentric lenses whose image angles overlap to such an extent that a triangulation of height differences from several camera images of different cameras is possible for each substrate position of the linear scanning area, or by a defined multi-camera arrangement with cameras positioned obliquely to the substrate, which, while observing a Scheimpflug condition, allows a complete overlap of the image angles of two adjacent cameras and thus a triangulation for each substrate position of the linear scanning area from only two camera images.

[0041] For the purposes of the present invention, definitions from the field of photography are used for the angle of view. The angle of view is understood to be the angle in the object space that is limited by the edges of the recording format of a camera (here, the camera of the recording unit). Accordingly, the angle of view is determined by the height and width of the recording format (as opposed to the diagonal, which is also frequently used and which specifies the maximum angle of view regardless of the actual aspect ratio of the recording format). The recording format is predetermined by the sensor format, whereby an object-side field of view (FOV) is defined via the lens image as the object space spanned by the horizontal and vertical angle of view.

[0042] Apart from the image format - height H and width B of the recording format - the angle of view is essentially only determined by the current focal length f of the lens. The focal length f However, it can only be used directly to define the angle of view when the lens is set to "infinity" (object-side telecentric lens). When imaging objects at a finite distance (short object distance), the image distance b larger than the focal length f and replaces it, which means that the horizontal angle of view for the width B of the recording format is α = 2 ⋅ arctan B / 2 ⋅ b results.

[0043] When using line scan cameras, the horizontal angle of view according to equation (1) is the decisive angle of view of the camera due to the almost linear sensor format and can therefore be used alone to define the linear object-side field of view (FOV).

[0044] The height deviations of the substrate along a scanning line determined by local triangulation with entocentric cameras can be used not only for the precise two-dimensional registration of targets but also for the precise tracking of the focus position of the processing beam moving linearly along a processing path. Tracking the processing beam along the processing path for the usual alignment of the exposure pattern to the registered target positions can be supplemented by adjusting mechanisms for quickly changing the focus of the processing beam based on the detected height fluctuations. The focus change must be performed at a frequency at least twice to ten times the usual scanning frequency of the processing beam (between 0.5 and 1 kHz for polygon scanners) and must therefore be controllable by simple linear position changes or changes in the radii of curvature of lenses or mirrors.

[0045] The invention provides a new possibility for improved exposure control in direct exposure processes for two-dimensional structures in photosensitive layers on printed circuit boards or wafers, which allows registration of targets "on the fly" by means of a linear scanning range independent of fixed location specifications of the targets for defined exposure fields and avoids the use of expensive telecentric lenses, as well as enabling flexible alignment and adaptation of the exposure pattern even for detected unevenness of the substrate.

[0046] The invention is explained in more detail below using exemplary embodiments and illustrations. In the following: Fig. 1: a schematic representation of a registration unit for target detection, designed as a multi-camera configuration, which includes a linear arrangement of entocentric cameras with overlapping image areas to achieve a linear scanning area across the entire object width transverse to the object movement; Fig. 2: a schematic representation of the registration unit, designed as a multi-camera configuration, in which the linear arrangement of the cameras is formed with pairs of entocentric cameras tilted relative to each other, which satisfy a Scheimpflug condition and in pairs have a completely overlapping image area; Fig. 3: a schematic representation for realizing the alignment of an entocentric camera under Scheimpflug conditions; Fig.Fig. 4: a schematic perspective view of two selected, consecutive line scans of the registration unit, as well as a result representation of the height variation of the object along two selected line scans; Fig. 5: a schematic view of the triangulation method used for height measurement in the overlap area of ​​two adjacent entocentric cameras; Fig. 6: a schematic view of the invention in a side view transverse to the direction of object movement with an enlarged view of the height fluctuation of the substrate surface, wherein the registration unit is schematically connected to the processing unit via a computer unit for converting the height measurement performed by triangulation into a precise control of the focusing within the processing beam scanned orthogonally to the drawing plane; Fig.Fig. 7: a schematic perspective view of the invention with registration unit and processing unit transverse to the substrate movement direction on an endless substrate that is tightly guided on a roller table system; Fig. 8: a schematic view of the implementation of the focus control for current focus tracking by means of a movable objective lens; Fig. 9: a schematic view of the implementation of the focus control for current focus tracking by means of a fixed angle mirror and a movable retroreflector; Fig. 10: a perspective view of the implementation of an elastically bendable mirror element that changes the mirror curvature using a linear actuator; Fig. 11: a schematic view of the bendable mirror element of . Fig. 10 ; Fig. 12: a schematic representation of the realization of the focus tracking with one elastic mirror element each according to Fig. 10 for mirror control separately in the direction of movement and in the transverse direction of the substrate.

[0047] The registration unit 1 according to the invention contains in an advantageous basic variant according to Fig. 1 a plurality of cameras 11 forming a linear configuration such that their optical axes 111 are aligned within the plane of the substrate 2 onto a line (scanning line 23 - only in Fig. 2 , Fig. 4 und Fig. 6 shown) are directed transversely to the direction of movement of the substrate 2 being moved past, and the substrate width being passed through is completely covered by partially overlapping image angles 112 or line-shaped fields of view (FOV) of the cameras 11. This is possible due to entocentric rather than telecentric lenses 15, with the cameras 11 also being equipped as line-scan cameras with one or a few parallel sensor lines 114.

[0048] The structure of cameras 11 with individual sensor lines 114 (so-called line cameras, in Fig. 1 not shown) is scalable, ie by combining several cameras 11, any required scanning width can be achieved. Since the exact position of target marks 22 can only be determined in the overlapping area 13 of the image angles 112 of two cameras 11, areas without overlap are lost for an exact measurement, so that the image angles 112 of the cameras 11 in the edge areas of the substrate 2 - if the overlapping areas 13 are limited to half the image angle 112 - can only be used halfway. For the cameras 11 in the edge area of ​​the substrate 2, Fig. 1 only half the angle of view 112 is shown.

[0049] The registration unit 1 is arranged with its elongated housing 12 transverse to the direction of movement of the target marks 22 (only in Fig. 2 and Fig. 7 drawn) of the substrate 2 to be scanned, wherein the substrate 2 is expediently guided underneath on a table system 3 (shown only as a support surface). As shown in the side view of Fig. 1 As can be seen, the recording unit 1 has intensive illumination means at a short distance from the substrate surface 21. These are mounted as light sources 17 in such a way that their illumination light only illuminates the substrate 2 without emitting direct light in the direction of the cameras 11, and the illumination can occur at various oblique angles of incidence (dark-field illumination) and, if necessary, with different spectral colors. Alternatively, it is also possible for the light from the light sources 17 to be coupled directly into the recording beam path of the cameras 11 as bright-field illumination (not shown).

[0050] The Fig. 1 The selected camera configuration consists of five cameras 11 with parallel optical axes 111 and relatively large overlapping areas 13 of the cameras 11, which correspond to half the image angle 112 of each camera 11. This achieves a gapless, almost linear scanning over the full width of the substrate 2, so that with a transversely moved substrate 2, as in the right side view of Fig. 1 As indicated by the arrow, it is possible to capture all target marks 22 located along the substrate width with one scan without mechanical camera movement. A virtually linear scanning by the registration unit 1 is realized by equipping it with cameras 11 in the form of line scan cameras. In order to generate a linear scanning area transverse to the direction of movement of the substrate 2, the entocentric cameras 11 must be aligned with optical axes 111 lying in a plane (not shown) onto a scanning line 23 (only in Fig. 2 drawn) onto the substrate surface 21 in such a way that their image angles 112 overlap there and form the overlapping areas 13 along the scanning line 23.

[0051] For typical substrate widths of 500-635 mm, entocentric cameras 11 (in the form of line scan cameras) with a scanning length of 330 mm in the focus area can be used for the currently required resolution (of 8-12 µm / pixel on the substrate 2), so that a registration unit 1 equipped with five cameras 11 can completely span the entire substrate width with a distance of 165 mm between the optical axes 111 of the cameras 11 in the direction of the substrate width (transverse extent of the substrate 2), whereby each location of the scanning line 23 is recorded simultaneously and redundantly in two different camera images of adjacent cameras 11.For this example, with the five cameras 11 directed onto the scanning line 23, a total scanning length of 660 mm results, which can be further moved closer together to ensure a gapless double scanning even in the case of positioning or mounting tolerances of the cameras 11 and / or height fluctuations Δz of the substrate 2, because the currently maximum substrate width (635 mm) is exceeded by 25 mm and thus additional small overlaps 14 of the overlapping areas 13 as well as reliable scanning beyond the edge areas of the substrate 2 are possible.

[0052] With the described arrangement of the cameras 11 according to Fig. 1 An actual linear scanning area can be scanned along the scanning line 23, which has a length-to-width ratio of more than 2,000. In general, the registration unit 1 can have a scanning line 23 with a length-to-width ratio of several thousand to 100,000, wherein the scanning width in the direction of movement y of the substrate 2 can be adjusted by electronic control (synchronization) of the readout modes and speeds of the cameras 11 by the computer unit 5.

[0053] The according to Fig. 1 The cameras 11 arranged with mutually parallel optical axes 111 are arranged along the scanning line 23 (only in Fig. 2 , 4 und 6 drawn) are arranged so close to one another that adjacent cameras 11 form an overlapping area 13 of at least half the image angle 112 and a further only slight overlap 14 with the image angle 112 of the next but one camera 11 is present if, with three or more cameras 11, a gapless coverage of the substrate surface 21 with the overlapping areas 13 is to be guaranteed even in the case of height fluctuations Δz of the substrate 2 and mechanical adjustment inaccuracies.

[0054] At the edges of the substrate 2 or the scanning line 23 formed by the cameras 11, only half the image angle 112 is usable due to the optical axes 111 of the cameras 11 being aligned perpendicularly to the substrate 2, so that an overlap area 13 is always present in the edge area of ​​the substrate 2. This is necessary because with entocentric cameras 11, the location of a target mark 22 present on the substrate 2 - the further this is spaced from the optical axis 111 - in the camera image depends very sensitively on the distance of the substrate surface 21 to the focal plane F n (only in Fig. 5 (referred to as) of the respective camera 11. To determine the distance of a target mark 22 positioned arbitrarily on the surface 21 of the substrate 2, a triangulation of two camera images from adjacent cameras 11 along the scan line 23 is therefore carried out and can also be continued for any other point on the scan line 23, provided that - apart from the target marks 22 - evaluable structures are present on the substrate 2.

[0055] In the z-direction, the exact determination of the height fluctuations Δz of the substrate surface 21 is limited to areas with detectable points (evaluable structures), so that the height profile of the substrate 2 may have to be supplemented by interpolated values.

[0056] The triangulation of the distance of the substrate surface 21 is crucial for determining the scanning location of a target mark 22, since any height fluctuations Δz of the surface 21 of the substrate 2 can lead to significant measurement errors in determining the x and y coordinates of a target mark 22, the closer it comes to the edge of the field of view 112 of the camera 11.

[0057] Light sources 17 positioned at a short distance from the substrate 2 in the housing 12 of the registration unit 1 and arranged at different oblique angles of incidence are provided for the targeted illumination of the scanning line 23, which is defined by the points of incidence of the optical axes 111 of the cameras 11.

[0058] As already mentioned above, the camera arrangement can be Fig. 1 One half of the image recording area (of the image angle 112) is not used by the outer cameras 11. By an alternative system design based on a Scheimpflug principle, which is Fig. 2 As shown, the above disadvantages can be avoided and the scanning areas within the entire image angle 112 of each of the adjacent, mutually inclined cameras 11 can be optimally utilized. However, the demands on the lenses 15 and the alignment and adjustment of the cameras 11 increase.

[0059] While a first seamless registration principle according to Fig. 1 for a total detection width of > 635 mm (> 25") with five cameras 11, the same detection width of the recording unit 1 can be achieved with the arrangement principle according to Fig. 2 scanned by only four cameras 11.

[0060] The detection width across all cameras 11 is dimensioned such that it is larger than the maximum processing area of ​​the processing unit 4 (only in Fig. 6 and Fig. 7 ), ie as the maximum length of the processing path 41 on the substrate 2, for example generated by a processing beam 45 scanned by means of a polygon scanner (only in Fig. 6 and Fig. 7 drawn).

[0061] A second constraint for the number and arrangement of cameras 11 is the required object-side optical resolution, which, depending on the target size to be detected, is often set at approximately 10 µm / pixel and, in the proposed examples, is realized at approximately 11 µm / pixel. A third constraint concerns the scanning speed, which should be adjusted to the desired throughput of printed circuit boards at a substrate speed of 1,000 mm / s to 1,800 mm / s.

[0062] As a compromise between high readout speed, maximizing the scanning width based on a large line length, and a reasonable price for the sensor lines 114, line scan cameras with sensor lines 114 that have more than 3,000 pixels and an edge length of 11 µm x 11 µm are primarily used for the cameras 11. The required width of the substrate 2 is adapted to the resolution of the line scan camera by means of the image scale of the lenses 15 of the cameras 11.

[0063] Fig. 2 shows in contrast to Fig. 1 a configuration of four cameras 11, which have mutually inclined optical axes 111 in a uniform camera plane along the scanning line 23 in order to cover the same scanning area as in Fig. 1 (635 mm) to be able to register the target marks 22 without gaps. Two cameras 11 form a pair of cameras 11 positioned under Scheimpflug conditions, which have a larger overlap area 13 of their image angles 112, which preferably corresponds to a complete overlap of the image angles 112 of the two cameras 11.

[0064] If the substrate 2 is not wider than this scan line 23 from the image angles 112 of the two neighboring cameras 11, the triangulation for each substrate point along the scan line 23 can be calculated from the two camera scans of just two neighboring cameras 11 tilted towards each other. Otherwise, if the substrate width is greater, further cameras 11 tilted towards each other in pairs can be lined up along the desired scan line 23 until the overlapping areas 13 of the two cameras 11 in each case cover the full width of the substrate 2. The overlapping areas 13 must at least touch each other, but should not overlap due to possible height fluctuations Δz of the substrate 2 and mechanical assembly orAdjustment tolerances of the cameras 11 also have a smaller overlap 14, which for maximum height fluctuations Δz and mounting tolerances of the cameras 11 always ensures the gapless scanning of the scanning line 23 on the substrate 2 by an additional overlap 14 of the overlap areas 13 formed by the complete overlap of the image angles 112 of the respective pairs of cameras 11. The advantage of this camera constellation according to . Fig. 2 lies in the fact that in the simplest case, exactly two cameras 11 "see" the same area of ​​the scanning line 23 while observing the Scheimpflug conditions and no parts of the image angles 112 of the cameras 11 remain unused. Fig. 1 This saves one camera 11 with the same length of the scanning line 23, ie with the same substrate width.

[0065] At the same time, with each pair of cameras 11 arranged under Scheimpflug conditions, an additional height triangulation for determining the height deviations Δz of the substrate surface 21 is possible if a sufficient number of target marks 22 or other scannable structures are present in the overlap area 13.

[0066] Fig. 3 represents one of the two adjacent cameras 11 inclined towards each other, in which the image of the object and image plane is rectified by setting different angles of inclination of the lens 15 and the sensor chip 113 (or inclination from lens to camera if the latter is viewed separately from the lens 15) and satisfying a Scheimpflug condition.

[0067] In Fig. 4 A sensor line 114 is shown schematically as a representative of a camera 11 of the recording unit 1 in order to illustrate the problem of a fluctuating height of the substrate surface 21 using the example of a substrate waviness. The upper right part of Fig. 4 shows schematically a table system 3 moved in the y-direction on which a substrate 2 is placed. Above the substrate 2 is the registration unit 1 reduced to a sensor line 114, which registers target marks 22 (only in Fig. 2 and Fig. 7 shown). For a substrate 2 to be scanned with respect to target marks 22, it is assumed or known that it is either unstable or wavy, resulting in height fluctuations Δz of the substrate surface 21. With a progressive relative movement of the sensor line 114 in the y-direction, different height profiles are recorded along the scanning line 23 (in the x-direction), which, depending on the position of the target marks 22 in the angle of view 112 of the entocentric cameras 11, have a reducing effect on the accuracy of the xy position determination of the target marks 22.

[0068] As a result of the feed of the table system 3, the sensor line 114 first scans the dotted scanning line 23', resulting in the dashed profile line in the diagram below. A few readout steps later, the sensor line 114 then detects the solid scanning line 23 and records a solid profile line that differs significantly from the dashed profile line. These locally very different height fluctuations Δz can be detected during processing by the processing unit 4 (only in Fig. 6 drawn) can lead to significant deviations in the exposed structures caused by defocusing (broadening) of the processing beam 45. The defocusing can only be corrected by refocusing if the locations of the height fluctuations Δz are precisely measured and the focusing of the processing beam 45 is adjusted along a processing line 41 adapted to the positions of the target marks 22 and the height fluctuations Δz.

[0069] In Fig. 5 An example of determining the height fluctuations Δz of the substrate surface 21 of the substrate 2 by means of triangulation is shown, in which two adjacent cameras 11 have parallel optical axes 111 and two focal planes F n and F n+1 of the sensor chips 113 lying in the same plane (shown here for n = 0), and an overlap region 13 of the image angles 112 (i.e., the linear scanning ranges) of two cameras 11 occurs. This is necessary because the detected target mark positions in the xy plane on the substrate surface 21 when imaged using entocentric lenses 15 are sensitively dependent on the object position in the z direction.Therefore, the target marks 22 to be registered are captured using camera images from a pair of adjacent cameras 11, and the target mark positions x, z are calculated using a triangulation method in which the height fluctuation for the z-direction is related to two reference planes, which are designated as substrate surface 21 in the lower plane and substrate surface 21' in the upper plane to indicate the surface change. The resulting x-position and the resulting height z are ultimately determined from the positions of the various calibration values ​​detected in both cameras 11. z 1 , z 0 with respect to the calibration planes of the substrate surfaces 21 and 21' is determined as follows. x p = x 1 T x 2 B − x 1 B x 2 T x 1 T − x 1 B + x 2 B − x 2 T z p = x 2 B − x 1 B x 1 T − x 1 B + x 2 B − x 2 T ⋅ z 1 − z 2

[0070] This means that in addition to the exact determination of the target position in x, the z-position in relation to the calibration planes z 1 , z0 can be determined. This makes an (absolute) height measurement possible.

[0071] In Fig. 6 is a side view of the registration unit 1 analogous to the right sectional view of Fig. 1 shown. and the linking of the determined position data of targets 22 (only in Fig. 2 and Fig. 7 drawn) via a computer unit 5 with the processing unit 4 is schematically shown. The problem of an uneven substrate 2 is stylized and enlarged for a wavy substrate surface 21, whereby the table system 3 was assumed to be a precision table.

[0072] When the table system 3 is moved in the y-direction, the registration unit 1 detects the x- and y-positions of target marks 22 arranged anywhere on the substrate 2 in a linear scanning area (scanning line 23, which is formed by several line cameras orthogonal to the drawing plane) by successive line scans. Due to the overlapping areas 13 formed by the cameras 11, the image angle 112 (only in Fig. 1 and Fig. 2 visible), a double scanning of each substrate point by two adjacent cameras 11 ensures that not only the exact target mark positions can be determined by means of triangulation, but also the height fluctuations Δz of the substrate surface 21 can be calculated. From the measured values ​​of the actual substrate height z in each point of the scanning line 23 that can be detected by imaged structures, the computer unit 5 then calculates, in addition to the usual alignment of the two-dimensional structure of the processing pattern with respect to the actual position of the target marks 22, the processing focus FP (only in Fig. 8 und 9 drawn) of the processing beam 45 along the processing path 41 is adjusted to the height fluctuations Δz of the substrate surface 21 by means of focus tracking 43.

[0073] In the Fig. 7 In the exemplary embodiment of the invention shown, the substrate 2 is a continuous substrate that is tautly guided from roll to roll (not shown) via a roller table system 31. The roller table system 31 can have a drum diameter between 200 and 500 mm.

[0074] Due to the tight material guide of the endless substrate 2, the scanning line 23 (in Fig. 7 not visible) of the cameras 11 of the registration unit 1 in front of the contact line of the substrate 2 with the roller table system 31.

[0075] The registration unit 1 is configured with two entocentric cameras 11 such that the scanning areas of the two cameras 11 form a scanning line 23 (not visible) that extends beyond the edges of the substrate 2 in order to also detect calibration marks 32 on the roller table system 31. For the necessity and handling of the calibration marks 32, reference is made to the unpublished DE 10 2019 128 198.9.

[0076] Otherwise, the registration unit 1 detects, as for flat substrates 2 Fig. 1 and 2 described, all target marks 22 located on the substrate 2 moved past with the roller table system 31, regardless of where they are located on the substrate 2.

[0077] Due to the special feature that the substrate 2 is guided as an endless substrate tautly on the roller table system 31 and thus does not have any height fluctuations Δz caused by waviness of the substrate 2, in this embodiment of the invention a local point-by-point height measurement can be dispensed with and the height measurement of the substrate surface 21 can be limited to a small overlap area 13 of the image angles 112 of the two cameras 11. In the overlap area 13, which in this case can be much smaller than half the image angle 112 of the two cameras 11, but at least 1 / 50 (>5 mm) of the detection range of the registration unit 1 (ie substrate width plus edge areas of the roller table system 31), preferably between 1 / 40 and 1 / 10, particularly preferably from 1 / 35 to 1 / 25 (approx. 10-15 mm), in analogy to the description of Fig. 5 - a triangulation calculation is performed, from which, in this example, the thickness of the substrate 2 and, if applicable, thickness variations over the entire length of the continuous substrate can be determined. A rapid focus change along the processing path 41 by means of the focus tracking 43, as in Fig. 6 described, but is generally not required. Nevertheless, the thickness information, i.e. the z-measurement value as the height of the substrate surface 21, is absolutely necessary for calculating the positions of the target marks 22 in the x and y directions. However, after an initial one-time determination (and, if necessary, repeated individual measurements from time to time), it can be retained for the calculations of the positions of all target marks 22 of the entire continuous substrate. The processing unit 4, which is aligned with the substrate surface 21 in a different radial plane of the roller table system 31, sends a scanned processing beam 45 for imprinting two-dimensional structures onto the processing path 41 of the substrate surface 21. However, it can also be arranged in a common plane (e.g. axial plane of the roller table system 31) diametrically opposite the registration unit 1 on the roller table system 31 (not shown).

[0078] For the height fluctuations Δz resulting from the unevenness of the substrate 2 or the substrate surface 21, as they lead to Fig. 4 und 6 As described above, according to the invention, a fast, point-like focus tracking 43 is required for the processing beam 45 along the linear processing path 41, so that the latter degenerates into a height-variable processing line.

[0079] Conventional autofocus systems are unsuitable for these rapid focus changes in the z-direction of the processing beam 45 scanned in the x-direction, which is a laser beam for photolithographic direct exposure processes. The basis for the rapid implementation of a local focus change along the processing path 41 according to the invention lies in the detection of the height fluctuations Δz of the substrate 2 simultaneously with the registration of the target marks 22 through redundant dual image acquisition using entocentric cameras 11 with seamless overlap areas 13 of their image angles 112 along a linear scanning area (scanning line 23).By means of this height measurement for each pixel of the scanning line 23, which is carried out by triangulation calculations upstream of the processing unit 4 in terms of time and location, the computer unit 5 can, in addition to the already usual adaptation of the data of two-dimensional structural patterns broken down into processing paths 41 to the position of the substrate 2 detected by the detected target marks 22, additionally calculate a point-by-point change in the processing focus FP along the processing path 41 of the processing beam 45 as a function of the height fluctuations Δz detected along the scanning line 23 of the registration unit 1 and calculated in the computer unit 5.

[0080] For the rapid focus changes along the processing path 41, additional means are required to supplement the otherwise conventional focusing optics 44. These must be of such quality that they exceed the scanning frequency of the processing beam 45 in the direction x transverse to the direction of movement y of the substrate 2 by at least twice. Preferably, the frequency of the focus change is between twice and thirty times, particularly preferably between five and twenty times, the scanning frequency of the processing beam 45, which, when using a polygon scanner, is in the range between 0.5 and 1 kHz.

[0081] Fig. 8 shows a first possibility for the realization of the focus tracking 43 by means of a movable lens 431.

[0082] Another variant of the focus tracking 43 according to Fig. 9 provides a fixed corner mirror 432 and a movable retroreflector 433, wherein the corner mirror 432 arranged in the already focused beam decouples the focused beam onto the retroreflector 433 and couples it back in after its reflection, and the retroreflector 433 moves towards or away from the corner mirror 432 in order to shift the focus FP in the z-direction.

[0083] Another mirror-based version of the focus tracking 43 is shown in the Fig. 10 bis 12 described. To do this, Fig. 10 und 11 a mirror assembly with a mirror of variable curvature, which is referred to below as elastically bendable mirror 434. For this purpose, Fig. 11 The operating principle is shown as a mechanical equivalent circuit diagram, in which the flexible mirror 434 is movably hinged in the edge region to a mirror holder 436 of a base body and is in contact in the central region with a linear actuator in the form of a piezo stack 435, which is supported on the same base body as the mirror holder 436. Fig. 10 shows the physical implementation of the mirror assembly as a nearly monolithic component, in which the mirror mount 436, a rectangular cuboid body, supports the elastically bendable mirror 434 as a monolithically thinned, pre-curved metal plate, which is elastically mounted on two parallel linear solid-state bearings. Thus, the bendable mirror 434 is a convex or concave cylindrical mirror that can only change its curvature in one spatial direction.

[0084] According to the execution of Fig. 12 For focus control, two crossed, e.g. in x and y direction, aligned elastically bendable mirrors 434 are arranged in a folded beam path of a multi-element focusing optics 44 for the processing beam 45 (only in Fig. 7 and Fig. 8drawn). In this case, the fast focus tracking 43 is integrated into the usual focusing optics 44, wherein the intermediate image plane 441 specifies the point-by-point generation of the two-dimensional structure (not shown) specified for processing, which is locally adapted to the position of the target marks 22 and which is transferred in the image plane 442 to the advantageously used polygon scanner (not shown) for guiding the processing beam 45 along the height-controlled processing line 41 onto the substrate 2. An advantage of this embodiment of the fast focus control with two one-dimensionally effective focus tracking 43 is that it is possible to work with different focuses in the x-direction (scanning direction of the processing beam 45) and y-direction (substrate movement direction).Thus, in the scan and cross-scan direction of the processing beam 45 (if necessary), different focus changes can be accommodated for astigmatic or other aspheric optics.

[0085] The invention makes it possible to realize a continuous and approximately one-dimensional scan line 23 on substrates 2 with height changes or curvatures by using entocentric cameras 11 for target registration, which not only allows detection of the targets 22 but also allows measurement of the height and height fluctuation Δz of the substrate 2. Based on a double scan with two differently positioned cameras 11 along the scan line 23 of the registration unit 1, triangulation measurements and calculations can be performed in the overlapping areas 13 using entocentric line cameras whose field of view 112 overlap. This allows for a height measurement.By means of point-by-point height measurements, height fluctuations Δz of the substrate surface 21 in the processing unit 4, which is also operated with a linear processing path 41, can be taken into account by means of a fast focus control, and can be corrected when controlling the processing beam 45 by means of a focus tracking 43 or can be turned into a height-adapted processing line at certain points along the processing path 41.

[0086] In addition, the dynamic focus adjustment along the processing path 41 also enables a correction of known imaging-related focus deviations (resulting from the optics design and optics manufacturing) of the processing beam 45 scanned along the processing path 41. Reference symbol

[0087] 1Registration unit 11Camera 111Optical axis 112Angle of view 113Sensor chip 114Sensor line 12Housing (of the registration unit) 13Overlap area (of the angle of view) 14Slight overlap (of angles of view) 15Lens (of camera 11) 151Lens adapter 16Image plane (of registration unit 1) 161Camera adapter 162Surface normal (of sensor chip 113) 17Light sources (for illuminating the linear scanning area) 2Substrate 21, 21'Substrate surface 22Target 23, 23'Scanning line 3Table system 31Roller table system (for continuous substrate) 32Calibration mark 4 Processing unit 41 Processing path 43 Focus tracking 431 Movable lens 432 (Fixed) angle mirror 433 (Movable) retroreflector 434 (Elastic) bendable mirror 435 Piezo stack (linear actuator) 436 Mirror mount 437 Rigid joint 44 Focusing optics 441 Intermediate image 442 Image plane 5Computer unit F n , F n+1 Focal plane FPFocus (tracked) xTransverse direction (in the width of the substrate 2) yMovement direction (in the length of the substrate 2) zFocusing direction (in the height of the substrate 2) ΔzHeight fluctuation (of the substrate surface 21)

Claims

1. A device for exposure control during photolithographic direct exposure of two-dimensional structures in photosensitive coatings on a substrate, comprising a registration unit for registering target marks located on a substrate surface, a movable table system for supporting and defined one-dimensional movement of the substrate below the registration unit, a processing unit with a controllable linear processing path for photolithographic processing of the substrate by means of a processing beam for imprinting the two-dimensional structures, and a computer unit for controlling the alignment between the processing path and the substrate by means of local adaptation of the photolithographic processing depending on the position of the substrate determined by registered target marks, characterized in that - in the registration unit (1), a plurality of entocentric cameras (11) are arranged in a linear alignment transverse to the one-dimensional movement of the substrate (2) to form an uninterrupted linear scanning region (23) across a predetermined width of the substrate (2) and have extended view angles (112) in the direction of the linear scanning region (23), wherein the view angles (112) of adjacent entocentric cameras (11) along the linear scanning region (23) have an overlap region (13) in order to capture redundant images of the substrate (2) of the adjacent cameras (11) in the overlap region (13), and - the computer unit (5) has means for calculating the position of the target marks (22) from the redundant image recordings in the overlapping region (13) of the adjacent entocentric cameras (11) by additionally using a vertical position of the target marks (22) determined by triangulation of a distance of the substrate surface (21).

2. The device according to claim 1, characterized in that the registration unit (1) is equipped with a plurality of entocentric cameras (11) for generating the linearly continuous, uninterrupted sensor region such that view angles (112) of adjacent cameras (11) have an overlap region (13) which is at least as large as half the view angle (112), wherein the computer unit (5) is set up in such a way as to determine target marks (22) positioned arbitrarily across the width of the substrate (2) independently of the position of the target mark (22) within seamlessly successive overlap regions (13) of the view angles (112) of respectively adjacent entocentric cameras (11) by triangulating a distance at arbitrary positions on the substrate surface (21).

3. The device according to claim 1 or 2, characterized in that the cameras (11) are line cameras in order to form the linear scanning region (23) by cameras (11) with a long scanning length, in a narrow, uninterrupted manner and with overlapping regions over the entire width of the substrate (2).

4. The device according to any one of claims 1 to 3, characterized in that the cameras (11) are directed onto the substrate surface (21) with mutually parallel optical axes (111), wherein the overlap regions (13) of the view angles (112) of all cameras (11) are of the same size.

5. The device according to any one of claims 1 to 4, characterized in that two adjacent cameras (11) are directed onto the substrate surface (21) with mutually inclined optical axes (111), wherein the overlap region (13) of the view angles (112) of the mutually inclined cameras (11) is set such that the view angles (112) of the two cameras (11) completely overlap on the substrate surface (21), wherein optionally the overlap region (13), which is formed by pairs of mutually inclined cameras (11), is connected without interruptions to at least one further overlap region (13) until the overlap regions (13) have an extent which corresponds at least to the width of the substrate (2), wherein an overlap (14) can be provided between pairs of mutually inclined cameras (11) in order to ensure a gap-free scanning region (23) of the registration unit (1) for all permissible height fluctuations (Az) of the substrate surface (21).

6. The device according to claim 5, characterized in that the pairs of mutually inclined cameras (11) are arranged so that they are subject to a Scheimpflug condition.

7. The device according to any one of claims 1 to 6, characterized in that the computer unit (5) additionally has a control for rapid focus tracking (43) of the processing unit (4) along the processing path (41) as a function of height fluctuations (Az) of the substrate surface (21), which includes triangulation of target marks (22) or any imaged structures of the substrate surface (21) on the basis of the redundant image recordings in the overlapping region (13) of the adjacent entocentric cameras (11).

8. The device according to any one of claims 1 to 7, characterized in that the registration unit (1) for target mark detection has only two entocentric cameras (11) arranged on a scanning line (23) transversely to the direction of movement of the substrate (2) with an overlap region (13) of one hundredth to one third of the view angle (112) of the camera (11) when the substrate (2) is guided as a flexible endless substrate on a roller table system (31) tight and without height fluctuations (Az), wherein the overlap region (13) of the view angles (112) of the two cameras (11) is designed such that the triangulation for the purpose of precisely determining the thickness of the substrate (2) in the overlap region (13) of the view angles (112) on the roller table system (31) can be used and can be assumed to be constant for the entire width of the roller table system (31).

9. The device according to any one of claims 1 to 8, characterized in that the registration unit (1) is equipped with light sources (17) for illuminating the linear, uninterrupted scanning region, which are arranged uniformly distributed in the housing (12) to realize a scanning line (23) homogeneously illuminated by dark field or bright field illumination, wherein optionally the light sources (17) are configured for continuous illumination and have a device for controlling at least one property of brightness, angle of incidence or spectral range, or wherein optionally the light sources (17) are arranged for continuous illumination in order to enable image recording by controlling the integration time of the sensor line (114) by means of an electronic shutter principle.

10. The device according to any one of claims 1 to 9, characterized in that a focus tracking (43) for height fluctuations (Az) of the substrate (2) is integrated into the processing unit (4), with which height fluctuations (Az) determined by the computer unit (5) by means of triangulation from images redundantly recorded with two adjacent cameras (11) of the registration unit (1) can be adapted in real time by rapid focus correction for each pixel of the registration unit (1), wherein the focus tracking (43) can be controlled on the basis of changes in a lens or mirror position or curvature.

11. The device according to claim 10, characterized in that the focus tracking (43) is controllable on the basis of the change in a mirror curvature at least in the transverse direction x relative to the direction of movement y of the substrate (2).

12. The device according to claim 11, characterized in that the focus tracking (43) is separately controllable on the basis of changes in a mirror curvature in the direction of movement y of the substrate (2) and a mirror curvature in the transverse direction x.

13. The device according to any one of claims 10 to 12, characterized in that the focus tracking (43) can be controlled by changing the mirror curvature by means of a piezoelectric element (435), wherein, optionally, the focus tracking (43) based on changes in a lens or mirror position or curvature can also be used to correct imaging-related focus deviations of an upstream focusing optics (44) or other upstream optical elements resulting from the optics design or optics manufacture.

14. A method for exposure control in photolithographic direct exposure of two-dimensional structures in photosensitive coatings on a substrate, comprising the following steps: - arranging a plurality of entocentric cameras (11) to form an uninterrupted linear scanning region (23) transversely to a direction of movement of the substrate (2) in a registration unit (1) for detecting target marks (22) located on the substrate (2), wherein the entocentric cameras (11) have extended view angles (112) along the linear scanning region (23) with an overlap region (13) formed by adjacent entocentric cameras (11) in order to obtain redundant image recordings of the substrate (2) from adjacent cameras (11) in the overlap region (13), - moving the substrate (2) on a movable table system (3) in a defined one-dimensional movement below the registration unit (1), - providing a processing unit (4) for photolithographically producing the two-dimensional structures with a processing beam (45) which is controllable along a linear processing path (41), - detecting the spatial position of target marks (22) distributed arbitrarily over a predetermined width of the substrate (2) with respect to length, width and a height position during a single pass of the substrate (2) through the linear scanning region (23) of the registration unit (1), - determining the positions of the target marks (22) distributed arbitrarily across the width of the substrate (2) from the redundant image recordings in the overlapping region (13) of the adjacent entocentric cameras (11) with additional use of a height position of the target marks (22) determined by triangulation of a distance of the substrate surface (21) from the redundant image recordings of adjacent entocentric cameras (11), - calculating data for the alignment and local adjustment of the processing of the substrate (2) with two-dimensional structures for the processing unit (4) for controlling the processing beam (45) along the linear processing path (41) aligned transversely to the direction of movement of the substrate (2), and - controlling the alignment between the processing path (41) and the substrate (2) and locally adapting the photolithographic processing depending on the position of the substrate (2) determined by registered target marks (22).

15. The method according to claim 14, wherein the calculation of the spatial position of target marks (22) distributed arbitrarily across the width of the substrate (2) on the basis of a height position during a pass of the substrate (2) is extended to the triangulation of further detectable structures of the substrate (2) in the redundant images recorded in the overlap region (13), and a rapid focus adjustment of the focus (FP) of the processing beam (45) is carried out by means of a focus tracking (43) along a processing path (41) on the basis of a control of a lens or mirror position or a mirror curvature, wherein optionally the rapid focus adjustment of the focus (FP) of the processing beam (45) is carried out at a frequency at least two to thirty times higher than the conventional scanning frequency for the processing beam (45).