DE102010006607A1 - Method for inspecting structures of semiconductor devices - Google Patents

Abstract

The invention relates to a method for inspecting test substrates in a prober, which has a chuck for receiving the test substrate 2, an inspection unit with a camera for inspecting the test substrate 2, a movement device for moving the chuck and / or the inspection unit relative to one another, a control unit for controlling of the movement device and a display for displaying an image recorded by the camera. For the inspection, an overview image 1 is generated of the test substrate 2 by means of the camera, in that a plurality of individual images 4 are recorded in different positions of the test substrate 2 and combined to form the overview image 1 on the display. Each individual image 4 is stored together with the coordinates of the position in which it was recorded. For navigation on the test substrate, an individual image 4 is selected in the overview image 1, whereupon a position linked to the individual image 4 is approached by means of the movement device and an image of the section of the test substrate 2 then located in the field of view 6 of the inspection unit is generated.

Description

The invention relates generally to a method for inspecting test substrates by means of an inspection unit of a prober. In particular, it concerns the inspection of test substrates which are coarser than the field of view of the inspection unit, so that the area to be inspected must be positioned in the field of view by means of a movement device of the tester in order to inspect the entire test substrate or regions of the test substrate lying outside the field of view.

As is known, a wide variety of test substrates are tested for various properties or subjected to special tests. The test substrates can be present in various stages of production and integration. Thus, tests of semiconductor chips, hybrid components, micromechanical and micro-optical components and the like are performed, which are still in the wafer assembly or isolated or already integrated in more or less complex circuits. In the following, the component to be inspected is the test substrate. This can be both an entire wafer and its individual components or already isolated components, depending on the current inspection task. It is equally possible that first a single component of a wafer and then another is to inspect each time using the same inspection method once on the single component and the other on the entire wafer.

For testing and inspection of test substrates, test stations are used which comprise a chuck with a surface for receiving test substrates. The chuck is usually movable by means of a movement device at least in the X and Y directions. The inspection station also has an inspection unit, a camera and / or a microscope, with which the test substrates can be inspected. This is done z. B. before or during the contacting of the contact pads of the test substrates by probe tips.

In the case of manual inspection of structures of test substrates, only a part of the entire wafer or only a part of the test substrate is usually visible due to the limited field of view of the inspection unit. The size of the image depends on the one hand on the size and resolution of the lens used in the inspection unit and / or the sensor used. The larger the objective magnification, the smaller the region of the test substrate to be displayed in an image, generally referred to as the image field. In a partial representation of the test substrate, the inspection unit operator must be positioned relative to the chuck by the operator of the inspection station, the operator, so that the part of interest for the inspection is within their field of vision.

The positioning of the inspection unit relative to the chuck is often done solely by a movement of the chuck. Alternatively, however, probes are also known whose inspection unit is movable, so that the chuck and inspection unit are delivered to one another or only the inspection unit to the chuck. In the following, the description will be described by way of example with reference to the movement of the chuck alone. A limitation to such positioning for inspection purposes is not intended.

If structures to be inspected of one and the same test substrate are spatially far apart, they no longer fit into the image field of the inspection unit at the same time. The operator must manually approach the required positions. This process is time consuming and monotonous, especially with frequent repetition. Quick manual inspection of a test substrate is possible only insofar as the structures to be inspected are close to each other or within the image field.

The object of the invention is to expand the image field of an inspection unit of a prober and to improve the navigation on the test substrate during the inspection.

There is provided an inspection method with which the image field of the inspection unit is virtually expanded by creating a virtual overview image of the test substrate or at least part of it. The overview image is obtained by two-dimensional juxtaposition of images taken by a camera, so-called tiling, which were taken at different positions of the chuck. The individual tiles of the overview image can either be recorded in different chuck positions within the X-Y plane. However, they can also include Z-direction jumps, depending on the structure of the test substrate and the target of the inspection.

This overview image composed of individual images is a still image that is only updated by explicit user interaction. It serves as a reference for further navigation for the inspection unit, eg. As in the context of adjustment or inspection of the contact position of the probe tips. The size of the virtual Image field of the overview image is chosen so that can be roughly navigated with sufficient accuracy on a test substrate, which is greater than the imageable with the inspection unit image detail. The fine positioning then takes place in the live image, which is obtained after starting the position selected in the overview image.

To produce the overview image, the chuck positions of interest are approached. In each position, a single image is taken, assigned to this the respective X and Y coordinates of Chuckposition and possibly also Z coordinates and scaling data and stored the image in connection with these data in the memory of the control unit of the Probers. The recording of the individual images can be done in various ways. Thus, an automated recording with presettable positioning steps of the chuck within a previously, z. B. manual, limited range on the test substrate possible. This can be done in particular when regular structures on the test substrate are to be inspected, such. B. Kontaktpadreihen or Kontaktpadarrays. Also, manual positioning to capture each image or a combination of both are possible. The individual images to be photographed can then be completely directly next to one another to completely map the region of interest of the test substrate or only selected regions thereof.

The individual images with the associated data are referred to below as tiles. From these tiles, the overview image is assembled on the basis of criteria that are favorable for the respective inspection task. Most of these criteria will be due to the spatial relationship of the individual tiles.

Alternatively, however, functional relationships that result from the image contents can also be decisive for the juxtaposition of the individual. In any case, the tiles are assembled on the basis of the coordinates without consideration of the image content. It follows from the latter that no positioning errors in the desired chuck position or lens distortions that would require image recognition methods are taken into account.

An indication of the number of tiles per overview image is, in addition to the inspection task, also the memory requirement, whereby it must be taken into account that both the individual tiles and the overview image must be kept in parallel at least briefly and the overview image can also be stored in the full resolution. So it can be helpful to save the images only in black and white, because in the case of colored images, both the computational and associated time expenditure as well as the storage requirements multiply. A reduction of the memory requirement can alternatively or additionally also be effected by a downward scaling of the resolution of the individual images. This has no effect on the subsequent inspection, since the tiles are used only for navigation and the inspection takes place after the start of the coordinates associated with the tile on the then obtained image, the live image.

The overview image is displayed on a display of the control unit of the prober, preferably together with the current live image of the test substrate to be obtained with the inspection unit. By means of a cursor, either a mouse pointer, a crosshair or similar, the operator can select and activate each of the tiles in the overview image. Upon activation of a tile, a signal is generated in the control unit to the movement device, which causes the start of the chuck position in which the activated tile was taken.

In the case of this so-called static positioning, coordinates and data of the activated tile always relate to the area of the test substrate recorded for the creation of the overview image, regardless of how far away the current position is from the location of the tile.

In one embodiment of the invention, the so-called dynamic positioning, the coordinates of the tiles of the overview image are transferred to regularly recurring, comparable substructures of the test substrate by counting the step size of the repetition of the substructure to the coordinates of each tile of the overview image. Each coordinate of each tile is corrected by a multiple of the values in the X, Y and possibly also the Z direction, which define the position of two neighboring substructures to each other. Based on the comparison of the current position with the recording area of the overview image, the number of repetitions of the substructure that lie between the two positions and thus how often and in which directions the corrections of the X, Y and Z coordinates are to be applied. The recording area of the overview image is z. B. based on a reference tile to define.

In this way, both a repetition in one direction and in several directions must be considered. In addition, this extension of the navigation on a variety of comparable structures, eg. B. applicable to components in the wafer composite, even without image recognition. The current position can only be determined from the coordinate comparison if the increment of the repetitions is known or previously determined. This allows navigation in the environment of the live image and on the substructure depicted there, even if the inspection unit is outside the area of the overview image. Each time a tile is activated in the overview image, it is automatically converted into a position within the current substructure, ie in the live image. It is also possible to navigate around the live image, but this assumes that the overview image has captured a complete substructure.

The invention will be explained in more detail with reference to an embodiment. In the accompanying drawing shows

1 a summary image composed of four by four tiles;

2A and 2 B a real overview and live image with the image data according to 1 ;

3 a real overview image compared to 2A larger field of view;

4 a wafer with a grid of components as repeating partial structures, with spatial assignment of an overview image and substructures adjacent thereto;

5 a display of an overview image and a live image.

1 provides an overview picture 1 which is from a chip 2 In the exemplary embodiment, the test substrate to be inspected 2 , is created. The chip 2 is part of a wafer 3 that has a variety of chips 2 includes. Unless the wafer 3 or to inspect, the wafer would 3 according to the terminology just described alike a test substrate 2 be.

The 16 frames 4 be successively from the chip 2 taken for each frame 4 the wafer 3 is repositioned accordingly. Every single picture 4 is combined with the position coordinates at which the frame 4 was recorded, and the image resolution saved. The saved frames 4 become an overview picture with suitable software 1 and can be displayed to the operator of a prober on a display.

The overview picture 1 was created from four by four tiles whose frames overlap one another slightly over the entire chip 2 represent what through the reduced copy of the overview image 1 on the wafer 3 should be clarified in conjunction with the connecting lines.

The single pictures 4 each have a resolution of 2048 × 2048 pixels. The exact data can be found in the following table: resolution 6971 × 6964 Number of tiles 16 Resolution of the tiles 2048 × 2048 Image field of the live image 3 × 3 mm ² color mode S / W Size of a pixel 1.5 μm lens 5x

Based on the size of a live image of 3 by 3 mm, the theoretical tiling of the selected tiling is 12 by 12 mm ² . The difference to the actual image field size of 10 × 10 mm ² in the exemplary embodiment is due to the overlap 5 the frames 4 , which was made here with about 20%. The image information in overlapping image areas is either overlaid or exclusively combined, if necessary parameterized.

A real overview picture 1 with a comparable tiling and resolution is in 2A shown. The single pictures 4 almost seamlessly fit together. By means of a suitable overlap of the individual pictures 4 are edge effects on the single image 4 that z. B. affect the representation in the border area, minimizable. In the exemplary embodiment, the individual images 4 not scaled down, so the high resolution in the details of the overview image 1 preserved. The better representation in 2A because not all 16 frames are 4 shown.

2 B provides a detailed picture from the middle of the overview picture 1 according to 2A The structures of the test substrate 1 are clearly and sharply displayed.

In the embodiment according to 3 a much coarser image field was displayed in the overview image 1. It consists of 50 by 50 individual images 4 together and gives an image field size of 150 by 150 mm ² , in this case without overlap. To the total resolution of the overview picture 1 To keep the storage capacity and computing time as small as possible, the individual images were 4 scaled down a lot this time. The data can be found in the following table: resolution 6870 × 6876 Number of tiles 2500 Resolution of the tiles 135 × 135 Image field of the live image 3 × 3 mm ² color mode S / W Size of a pixel approx. 22 μm lens 5x

Due to the lower detail resolution in this embodiment, a single image would be used 4 only significantly blurred contours of the test substrate 2 result, but still sufficient to z. B. detect contact pads. With a correspondingly low structure resolution, sufficient storage capacity and computing time, a complete image of a wafer would also be possible 3 in the overview picture 1 possible.

4 shows a wafer according to 1 in which one of the chips 2 an overview picture 1 from two times two tiles 4 with an overlap 5 was created. In the embodiment according to 4 becomes this overview picture 1 through the dynamic positioning described above to others, not for the creation of the overview image 1 used chips 2 of the wafer 3 applied to navigation. For distinguishing the starting position, in which the overview image 1 was taken from the later positions where the same overview picture 1 only on the repetitive, comparably structured chips 2 on the wafer 3 is applied, the former position is surrounded by a solid line, whereas all other positions have a dashed border.

In the illustrated dynamic positioning of a chip 2 for inspection purposes, the wafer becomes 3 by means of the movement device of the chuck of the tester (not shown) relative to the stationary inspection unit (not shown), in this case a camera, insofar as the field of view is moved 6 the camera away from the area of the recorded overview image 1 lies. In the illustration according to 4 is for better clarity because of the field of vision 6 opposite the wafer 3 shown displaced.

After the overview picture 1 in the lower part of the wafer 3 taken and the wafer 3 so far was that the field of view 6 in the upper area of the wafer 3 Static positioning would lead to a selection and activation of one of the four tiles 4 of the overview picture 1 the wafer 3 automatically reset so that the activated tile 4 of the originally recorded chip 2 exactly in the field of vision 6 the camera is lying.

In dynamic positioning, the difference between the currently approached position and the field of view of the camera, whose X and Y coordinates are known, and the position coordinates of the activated tile, are used 4 determines how many chips 2 lie between these two positions in both the X and Y directions. From the knowledge of the size of each chip in the X and Y directions, the coordinates of the activated tile become 4 on those of the chip 2 converted, located in the immediate vicinity of the field of view 6 is located, ie where the largest overlap with the field of view 6 results. This largest overlap can z. B. can be determined by determining which chip 2 itself the center of the field of vision 6 located.

After the attacked chip 2 the tiles 4 of the overview picture 1 is reassigned, the corrected tile 4 this chip 2 approached. In this way, first roughly a region on the wafer 3 approached and there the overview picture 1 on the in the vicinity of the field of view 6 located chips 2 be used to navigate there. In 4 This dynamic positioning has been achieved in several distinctly spaced areas of a wafer 3 repeated.

5 the display presents an overview picture 1 and a live picture 7 on a display 10 , The overview picture 1 is composed of 4 tiles. Because the live picture 7 that is, which is in the current position with the camera recordable, it corresponds to the field of view of the camera according to 4 ,

In the overview picture 1 on the display is using a crosshair 11 , which serves as a cursor, the current field of view 6 the camera, ie the inspectable area of a test substrate 2 shown. In 5 This is an excerpt from a chip 2 , Accordingly, also presents the live picture 7 this section, but with a higher resolution. If another section is to be displayed, the overview screen appears 1 by moving the crosshair 11 navigates until the desired section in the crosshairs 11 lies. At this point you can click on it with the crosshairs 11 the tile 4 activated, which is located under the crosshair. As a result of the activation of the tile 4 The movement device of the prober (not shown) is activated and the chuck is positioned so that the activated tile 4 or a dynamically corrected tile 4 located in the field of view of the camera.

LIST OF REFERENCE NUMBERS

1

Overview screen

2

Test substrate, chip

3

wafer

4

Single picture, tile

5

overlap

6

field of view

10

Display, display

11

Cursor, crosshair

Claims (4)

Method for inspecting test substrates in a prober, which includes a chuck for holding the test substrate ( 2 ), an inspection unit with a camera for inspection of the test substrate ( 2 ), a movement device for moving the chuck and / or the inspection unit relative to one another, a control unit for controlling the movement device and a display for displaying an image recorded by means of the camera, comprising the following method steps: - holding the test substrate ( 2 ) on the contact surface of the chuck and position the test substrate ( 2 ) by means of the movement device in the field of vision ( 6 ) of the inspection unit, - generate an overview image ( 1 ) of at least a portion of the test substrate ( 2 ) by means of the camera by a plurality of individual images ( 4 ) of said section in different positions of the test substrate ( 2 ) relative to the field of view ( 6 ) of the inspection unit and on the display for the overview image ( 1 ), - save each frame ( 4 ) together with the coordinates of the position in which it was taken, - selecting a single image ( 4 ) in the overview screen ( 1 ), whereupon a control signal is generated in the control unit for starting up a frame (with 4 ) linked position by means of the movement device and - approaching this position and imaging of the position in this position in the field of view ( 6 ) section of the test substrate ( 2 ). The method of claim 1, wherein the coordinates of the one frame ( 4 ) are determined by an integral multiple of a predefined step size and the number of correction steps from the distance determined between a position during the generation of the overview image (FIG. 1 ) and a currently set, deviating position. Method according to one of the preceding claims, wherein the coordinates of the position of each individual image ( 4 ) also the scaling of the individual images ( 4 ), in particular their resolution is stored. Method according to one of the preceding claims, wherein the overview image ( 1 ) a test substrate ( 2 ) completely maps.

Images