Automatic wafer alignment technique

• US 4,475,122 A
• Filed: 11/09/1981
• Issued: 10/02/1984
• Est. Priority Date: 11/09/1981
• Status: Expired due to Term

• Current Assignee: Interserv Corporation
• Original Assignee: TRE Semiconductor Equipment Corporation
• Inventors: Green, Lawrence S.
• Primary Examiner(s): Britton, Howard W.
• Application Number:
  US06/319,782
• Time in Patent Office: 1,058 Days
• Field of Search: 358/101, 358/106, 358/107, 382/44-48, 382/8, 364/468, 356/400, 356/401
• US Class Current: 348/87
• CPC Class Codes: G03F 9/7088 Alignment mark detection, e.g. TTR, TTL, off-axis detection, array detector, video detection

Abstract

This automatic wafer alignment technique may be used in a step-and-repeat photomicrolithographic exposure system accurately to prealign a wafer before exposure to the B-level and subsequent reticles, and automatically to align the wafer at each die site prior to exposure. A search technique optimizes location of the global targets used for wafer prealignment. The search begins at the most likely target position and proceeds through a search area established by the maximum expected rough prealignment error. A wafer alignment target configuration consisting of a cross with one elongated arm and a short crossbar is used to optimize target verification. To locate targets a video image is digitized and the average intensity at each video scan line and column is obtained and stored. This data is used to determine the presence in the video image of feature edges. A table of all such edges is established. These listed edges are correlated to determine all pairs of edges that may represent potential targets. From these, the best target is selected by comparing certain weighted parameters of each potential target.

Claims

1. A method for use in a system in which a video camera is used to view a target of expected characteristics, said method being used for ascertaining whether a feature evident in the video image corresponds to the actual target, said video camera providing output data consisting of an average intensity value for each scan line or column of said video image, comprising:

ascertaining, from said video data, each feature edge evident in the video display,

sequentially correlating the data for each ascertained edge with the data for each other ascertained edge to determine all pairs of edges which represent potential targets, and

selecting the best target by weighted comparison of selected parameters of all determined pairs of edges.

2. The method of claim 1 comprising the initial step of:

entering into a "data buffer" memory a list of the average intensity value for each scan line or column of said video image and the signed difference in intensity value between consecutive ones of said scan lines or columns, said ascertaining being accomplished by reference to said data buffer memory.

3. The method of claim 2 wherein said ascertaining comprises:

determining from the values stored in said data buffer memory the presence of each run of consecutive scan lines or columns for which the intensity differences between consecutive points all are of the same sign or zero, but in which no two or more consecutive intensity differences are zero, each such run representing an edge, and

establishing in an "edge buffer" memory a list of all such edges including the scan line or column at which each such edge run begins and ends, and of the total intensity difference between the beginning and end of each such edge run.

4. The method of claim 3 wherein said ascertaining further comprises:

determining for each edge run whether it is of the same sign as the immediately preceding edge, and if so, of further determining whether the beginning of said each edge run is separated from the end of said immediately preceding edge by less than a certain small number of scan lines or columns, and, if so,

listing said each edge run and said immediately preceding edge in said edge buffer memory as a single edge beginning at the starting point of said immediately preceding edge and ending at the end of said each edge run.

5. The method of claim 3 or 4 wherein said correlating to determine potential targets comprises:

determining those pairs of edges, listed in said edge buffer memory, for which the edges are of opposite sign and have run heights that are within a certain range of one another, and that are separated from one another by a width that is within a certain preselected range, and

establishing in a "target buffer" memory a list of all pairs of edges that meet such criteria, each such pair representing a potential target.

6. The method of claim 5 wherein said selecting comprises:

for each potential target listed in said target buffer memory, establishing a set of weighted criteria each consisting of a parameter selected from the group consisting of absolute intensity of the potential target, ratio of the two edge heights of said potential target, width of said potential target, and relationship of intensity of potential target to the intensity of the surrounding area of said video image, each such parameter being weighted by a corresponding weighting factor,

algebraically combining all of the weighted criteria for each potential target, and

selecting as the best target that potential target having the optimum value of combined weighted criteria.

7. A method for determing the location of edges of features evident in a video image sensed by a video camera, comprising:

obtaining for each video scan line or column a digital value representing the average intensity along such scan line or column,

comparing, for each such scan line or column, the intensity value of that scan line or column with the intensity value of each the next two successive scan lines or columns, thereby to obtain a first intensity difference between said each scan line or column and the next one following and a second intensity difference between said each scan line or column and the second next following one,

determining from (a) the intensity difference between said each scan line or column and the next preceding scan line or column, and from (b) said first and second intensity differences, that the preceding, each and next scan lines or columns belong to a potential edge, by using the criteria that, for a potential edge, the signs of such preceding and first intensity differences are the same, or that said first intensity difference is zero and the signs of the said preceding and second intensity differences are the same, and

determining the end point of such edge by the criteria that an edge ends when two consecutive intensity differences are zero, or that the sign of one intensity difference is the opposite of the sign of the preceding two intensity differences.

8. A method for use in a wafer alignment system employed with a wafer processing apparatus in which a semiconductor wafer is to be aligned in a certain orientation with respect to the reference axes of said system, said wafer containing at least one alignment target designated as a global alignment target, the wafer being roughly prealigned to within certain rotational error and lateral offset error tolerances, and in which a small region of the wafer is viewed by a video camera for alignment purposes, the method being used for searching for said global target, comprising:

first positioning said wafer so that said video camera views a first area thereon at which said global target would be located if the wafer were correctly positioned, and determining from the output of said video camera if the global target is visible at said area,

thereafter moving said wafer alternately in one direction and the opposite direction along the system axis closest to that along which the target would be displaced as a result of rotational error in rough positioning, by amounts increasing in each such direction by steps corresponding to the dimension in that direction of the region viewed by said video camera, such being repeated until the distance searched in each direction from said first area is substantially equal to the expected maximum offset of said target which would result if the rough alignment were to have said maximum expected rotational error, and determining from said vidio camera output, at each such location, if said global target is visible at such location, and

thereafter repeating such moving and determining first along one then along another of two paths parallel to the initial movement path but laterally offset therefrom by respective positive and negative amounts corresponding to the expected maximum lateral offset error of said wafer, and

terminating such search at the location at which said alignment target is visible by said video camera.

9. A method according to claim 8 wherein said wafer alignment target is of the shape of a cross having an elongated arm generally oriented in a direction normal to said closest system axis, and having a relatively shorter laterial crossbar near the middle of said elongated arm.

10. A method for use in an apparatus requiring alignment of one article, said one article containing an alignment target that is viewed by a video camers, said method being used for verication that a feature in the image being viewed by said video camera is actually said alignment target, comprising:

utilizing for the shape of said target a cross having a relatively elongated arm and a relatively short crossbar near the middle of said elongated arm,

initially positioning said one article so that the image viewed by said video camera is near said middle and includes part of said elongated arm and part of said crossbar, and, at such position, verifying from the output of said video camera the presence in said image of both elongated arm part and said crossbar part, and

thereafter successively moving said one article to positions at which the image viewed by said video camera would be expected to include only part of one end of said elongated arm, only part of the other end of said elongated arm, only the region beyond one end of said elongated arm and only the region beyond the other end of said elongated arm, and at each such position, verifying from the output of said video camera the expected presence or absence of said arm parts.

11. A method according to claim 10 wherein said one article is to be aligned with a second article, and wherein

said second article includes an alignment window through which said video camera views said alignment target, the length of said elongated arm on each side of said crossbar being longer than the width of the area viewed through said window by said video camera.