Inspection system for array of microcircuit dies having redundant circuit patterns

We claim:

1. In an imaging system that includes first and second lenses positioned along an optic axis, the first lens producing from a specimen a spatial frequency spectrum whose frequency components can be selectively filtered and the second lens producing an image of defects present in the specimen, a method of detecting defects in a specimen that includes an array of normally substantially identical dies, each of the dies having many redundant circuit patterns, comprising:

illuminating plural die circuit patterns;

generating a light pattern representing substantially the Fourier transform pattern of the illuminated die circuit patterns, the light pattern including intra-die interference pattern information;

positioning an optical filter to receive the light pattern and to block spatial frequency components thereof, the optical filter having relatively transparent and relatively nontransparent portions, the relatively nontransparent portion conforming to the Fourier transform pattern of an error-free reference pattern corresponding to the die circuit patterns;

collecting spatial frequency components not blocked by the optical filter to form an image of the defects, the collected spatial frequency components corresponding to a small number of die circuit patterns relative to the number of die circuit patterns in the array of dies and residing in a spatial region intercepting the optic axis; and

processing unblocked intra-die spatial frequency components to determine the location and size of a possible defect in the die.

2. The method of claim 1 which further comprises changing the position of the specimen relative to the position of the optic axis so that different ones of the die circuit patterns are positioned within the spatial region intercepted by the optic axis, thereby to process the intra-die spatial frequency components of the different ones of the die circuit patterns.

3. The method of claim 1 in which the processing of the unblocked intra-die spatial frequency components is accomplished by positioning a light sensitive detector surface generally centrally about the optic axis, the light sensitive detector surface having an area that is smaller than the surface area of the image of the defects.

4. The method of claim 1 in which the first and second lenses cooperate to receive light diffracted by, and provide an image from the spatial frequency components corresponding to, the illuminated die circuit patterns.

5. The method of claim 4 in which the first lens comprises a first lens section of plural elements and the second lens comprises a second lens section of plural elements, the first and second lens sections forming a near diffraction-limited lens system of asymmetric character.

6. The method of claim 1 in which the illuminating means emits nearly collimated light, the method further comprising:

defining with respect to the specimen plural adjacent stripes, each stripe including a series of adjacent dies;

moving the specimen and the collimated light relative to each other along the length of each stripe to illuminate the die circuit patterns in proximal position to the optic axis; and

processing the unblocked intra-die spatial frequency components corresponding to the die circuit patterns in proximal position to the optic axis.

7. The method of claim 6 in which the specimen is movable and the collimated light remains fixed relative to the optic axis.

8. The method of claim 1 in which the relatively transparent and relatively nontransparent portions of the optical filter are developed by computer generation techniques.

9. The method of claim 1 in which the relatively transparent and relatively nontransparent portions of the optical filter are developed by positioning a recording medium in the location of the Fourier transform pattern and exposing the recording medium to light propagating from the specimen.

10. The method of claim 1 in which the collected spatial frequency components correspond to fewer than all of the illuminated die circuit patterns.

11. In an imaging system that includes first and second lenses positioned along an optic axis, the first lens producing from a specimen a spatial frequency spectrum whose frequency components can be selectively filtered and the second lens producing an image of defects present in the specimen, a method of detecting defects in a specimen that includes an array of normally substantially identical dies occupying a first area of the specimen, each of the dies having many redundant circuit patterns, comprising:

illuminating a second area of the specimen, the second area containing die circuit patterns and intercepting the optic axis;

generating a light pattern representing substantially the Fourier transform pattern of the illuminated die circuit patterns, the light pattern including intra-die interference pattern information;

positioning an optical filter to receive the light pattern and to block spatial frequency components thereof, the optical filter having relatively transparent and relatively nontransparent portions, the relatively nontransparent portion conforming to the Fourier transform pattern of an error-free reference pattern corresponding to the die circuit patterns;

collecting spatial frequency components not blocked by the optical filter to form an image of the defects; and

processing only unblocked intra-die spatial frequency components to determine the location and size of a possible defect in the die.

12. The system of claim 11 in which the size of the first area differs from that of the second area.

13. The system of claim 12 in which the second area is substantially smaller than the first area.

14. The system of claim 12 in which the second area contains more than one die.

15. In an imaging system that includes first and second lenses positioned along an optic axis, the first lens producing from a specimen a spatial frequency spectrum whose frequency components can be selectively filtered and the second lens producing an image of defects present in the specimen, a method of detecting defects in a specimen that includes an array of normally substantially identical dies occupying a first area of the specimen, each of the dies having many redundant circuit patterns, comprising:

illuminating die circuit patterns included within a second area of the specimen;

generating a light pattern representing substantially the Fourier transform pattern of the illuminated die circuit patterns, the light pattern including intra-die interference pattern information;

positioning an optical filter to receive the light pattern and to block spatial frequency components thereof, the optical filter having relatively transparent and relatively nontransparent portions, the relatively nontransparent portion conforming to the Fourier transform pattern of an error-free reference pattern corresponding to the die circuit patterns;

collecting spatial frequency components not blocked by the optical filter to form an image of the defects, the collected spatial frequency components corresponding to fewer than all of the illuminated die circuit patterns; and

processing the unblocked intra-die spatial frequency components to determine the location and size of a possible defect in the die.

16. The method of claim 15 in which the second area is substantially smaller than the first area.

17. In an imaging system that includes first and second lenses positioned along an optic axis, the first lens producing from a specimen a spatial frequency spectrum whose frequency components can be selectively filtered and the second lens producing an image of defects present in the specimen, a method of detecting defects in a specimen that includes an array of normally substantially identical dies, each of the dies having many redundant circuit patterns, comprising:

illuminating plural dies circuit patterns;

generating a light pattern representing substantially the Fourier transform pattern of the illuminated die circuit patterns, the light pattern including intra-die interference pattern information;

positioning an optical filter receive the light pattern and to block spatial frequency components thereof, the optical filter having relatively transparent and relatively nontransparent portions, the relatively nontransparent portion conforming to the Fourier transform pattern of an error-free reference pattern corresponding to the die circuit patterns;

collecting spatial frequency components not blocked by the optical filter within a region proximal to the optic axis to form an image of the defects; and

processing unblocked intra-die spatial frequency components to determine the location and size of a possible defect in the die, the processed spatial frequency components corresponding to a small number of die circuit patterns relative to the number of die circuit patterns in the array of dies and lying in a spatial region intercepting the optic axis.

18. An optical system for detecting defects in a specimen pattern of a type that includes an array of normally essentially identical dies of which each has many redundant circuit patterns and which occupy a first area of the specimen, the system comprising:

illuminating means for illuminating a second area of the specimen, the second area being occupied by plural die circuit patterns;

pattern generating means for generating a light pattern representing substantially the Fourier transform pattern of the illuminated die circuit patterns, the light pattern including intra-die interference pattern information;

optical filter means receiving the light pattern for blocking spatial frequency components thereof, the optical filter means having relatively transparent and relatively nontransparent portions, the relatively nontransparent portion conforming to the Fourier transform of an error-free reference pattern corresponding to the die circuit patterns;

collecting means for collecting the spatial frequency components not blocked by the optical filter means; and

processing means for processing only the unblocked intra-die spatial frequency components to determine the location and size of a possible defect in the die.

19. The system of claim 18 in which the illuminating means emits nearly collimated light and which further comprises positioning means for changing the position of the specimen relative to the position of the collimated light so that different ones of the die circuit patterns occupy the second area of the specimen illuminated by the collimated light, thereby to process the intra-die spatial frequency components of the different ones of the die circuit patterns.

20. The system of claim 18 in which the pattern generating means and the collecting means comprise respective first and second lenses positioned along an optic axis that intersects the second area of the specimen illuminated by the illuminating means.

21. The system of claim 18 in which the pattern generating means and the collecting means comprise respective first and second lenses that cooperate to receive light diffracted by, and provide an image from the spatial frequency components corresponding to, the illuminated die circuit patterns.

22. The system of claim 21 in which the first lens comprises a first lens section of plural elements and the second lens comprises a second lens section of plural elements, the first and second lens sections forming a near diffraction-limited lens system of asymmetric character.

23. The system of claim 21 in which the first lens comprises a first lens section of plural elements and the second lens comprises a second lens section of plural elements, the first lens section forming the Fourier transform pattern and cooperating with the second lens section to provide a magnified image of the defects in the illuminated die circuit patterns.

24. The system of claim 18 in which the pattern generating means and the collecting means comprise a folded Fourier transform optical system that receives light diffracted by, and provides an image from the spatial frequency components corresponding to, the illuminated die circuit patterns.

25. The system of claim 24 in which the specimen comprises a semiconductor wafer.

26. The system of claim 24 in which the optical filter means comprises a liquid crystal layer.

27. The system of claim 26 in which the relatively nontransparent portion of the liquid crystal layer scatters light of the spatial frequencies incident to it.

28. The system of claim 18 in which the optical filter means comprises a liquid crystal layer.

29. The system of claim 28 in which the relatively nontransparent portion of the liquid crystal layer scatters light of the spatial frequencies incident to it.

30. The system of claim 18 in which the optical filter means comprises exposed light sensitive material.

31. The system of claim 18 in which the optical filter means comprises a lens assembly that has an aperture of at least .+-.15.degree..

32. The system of claim 18 in which the Fourier transform light pattern represents the Fourier transform image.

33. The system of claim 18 in which the specimen comprises a semiconductor wafer.

34. The system of claim 18 in which the second area is substantially smaller than the first area.

35. The system of claim 18 in which the illuminating means emits nearly collimated light and the processing means comprises a light sensitive detector having a light sensitive surface positioned generally centrally about the optic axis, the light detector including plural light detecting elements arranged in a first array of rows and columns and defining in the light pattern plural adjacent stripe regions each of which includes plural pixel elements arranged in a second array of rows and columns, and each light detecting element being operable to provide a measured energy value corresponding to the amount of light present in any one of the pixel elements, and the system further comprising;

positioning means for positioning the specimen relative to the collimated light to scan the light detecting means along a stripe region of the light pattern so that in succession each light detecting element in one column of the first array traverses and acquires an energy value corresponding to the amount of light present in a pixel element in one column of the second array;

accumulating means to accumulate a total energy value proportional to the sum of the energy values acquired for the pixel element by all of the light detecting elements in the one column of the first array; and

means to determine from the total energy value whether the amount of light in the pixel element represents a defect in the specimen subject.

36. The system of claim 35 in which the light detector comprises a charge-coupled device.

37. The system of claim 35 in which the collimated light remains stationary and the positioning means scans each one of the stripe regions across the light sensitive surface in a serial manner.

38. The system of claim 37 in which the positioning means continuously moves each stripe region across the collimated light.

39. The system of claim 35 in which the first array has a first row and N total number of rows and which further comprises position-detecting means for detecting the position of the first array relative to the stripe region, the position-detecting means cooperating with the accumulating means so that each one of the light detecting elements in the first row of the one column never accumulates more than one energy value for any one of the pixel elements of the second array with which it becomes aligned, and each one of the light detecting elements in the Nth row of the one column has accumulated N number of energy values for any one of the pixel elements with which it becomes aligned.

Description Submit all comments and votes

TECHNICAL FIELD

The present invention relates to inspection systems for use in the manufacture of microcircuits and, in particular, to a real-time defect inspection system for use in the manufacture of microcircuits of the type that includes an array of dies each having many redundant circuit patterns.

BACKGROUND OF THE INVENTION

Two exemplary and very similar inspection systems for pattern defects in photomasks employed in the large-scale manufacture of semiconductor devices and integrated circuits are described in U.S. Pat. Nos. 4,000,949 of Watkins and 3,614,232 of Mathisen. The systems of Watkins and Mathisen contemplate the simultaneous inspection of all of the dies on a photomask which contains a regular array of normally identical dies to detect the presence of nonperiodic defects, i.e., defects in one die not identically repeated in the remaining dies of the array.

This task is accomplished by illuminating simultaneously all of the dies of a specimen photomask with collimated coherent light emanating from a laser to develop a composite diffraction pattern whose spatial distribution is the combination of two components. The first component is the interference pattern of the array of dies, and the second component is the interference pattern of a single die of the array. The first and second components are sometimes called an inter-die interference pattern and an intra-die interference pattern, respectively. The light transmitted by the photomask strikes a double-convex lens which distributes the light on a spatial filter positioned a distance equal to one focal length behind the lens.

The spatial filter comprises a two-dimensional Fourier transform pattern of a known error-free reference photomask against which the specimen photomask is compared. The filter is opaque in the areas corresponding to spatial frequency components of the error-free Fourier transform pattern and is transparent in areas not included in the error-free Fourier transform pattern. Neither the Watkins patent nor the Mathisen patent specifies the design parameters of the lens. The Mathisen patent states only that the lens is of suitable numerical aperture and magnification power to cover the area of the specimen photomask.

The spatial frequency components corresponding to the defects in the specimen photomask are largely transmitted through the spatial filter and can be processed in either one of two ways. In the Watkins system, the light transmitted through the spatial filter strikes another double-convex lens that is properly positioned to define an image of the specimen photomask, absent any information blocked by the spatial filter. The imaging light not blocked by the spatial filter appears in locations that represent the position in the specimen photomask where defects are present. In the Mathisen system, the light transmitted through the spatial filter is sensed by a photodetector that produces an output signal which activates a "no-go" alarm.

The Watkins and Mathisen patents imply that systems of the type they describe require both inter- and intra-die interference pattern information to determine the presence of defects in the specimen pattern. The inter-die interference pattern information is of particular concern because it consists of very closely spaced light spots that are extremely difficult to resolve by a Fourier transform lens. The realization of such a lens is further complicated for inspection systems that use an inverse Fourier transform lens to form an image of the specimen pattern from the Fourier transform light pattern. The reason is that the design of each of the lenses is compromised to accomplish an overall system design that accomplishes both the Fourier transform pattern and image forming functions. It is, therefore, exceedingly difficult to obtain from such a system design the resolution required to acquire inter-die interference pattern information. The above lens design problem is encountered in systems of the type that simultaneously inspects the entire area of each of the dies of a specimen photomask array and, as a consequence, renders such systems unreliable and impracticable for commercial use.

SUMMARY OF THE INVENTION

An object of the present invention is, therefore, to provide a reliable defect inspection system for use in the manufacture of microcircuits.

Another object of this invention is to provide such a system that applies the techniques of Fourier optics but does not contemplate the use of inter-die interference pattern information to determine the presence of defects in the manufacture of microcircuits of the type that comprises an array of normally identical dies.

A further object of this invention is to provide such a system that is capable of developing from a microcircuit pattern an essentially aberration-free Fourier transform light pattern from which an accurate image corresponding to defects in the microcircuit pattern can be formed.

Still another object of this invention is to provide an inspection method that uses intra-die interference pattern information to determine the presence of defects in a microcircuit array pattern of normally identical dies.

The present invention relates to a method and system for use in the manufacture of microcircuits and is described herein by way of example only with reference to a real-time inspection system for defects in surfaces of semiconductor wafers of the type that includes an array of circuit dies of which each has many redundant circuit patterns. Such semiconductor wafers include, for example, random access and read only memory devices and digital multipliers.

Two preferred embodiments of the inspection system employ a Fourier transform lens and an inverse Fourier transform lens positioned along an optic axis to produce from an illuminated area of a patterned specimen wafer a spatial frequency spectrum whose frequency components can be selectively filtered to produce an image pattern of defects in the illuminated area of the wafer. The lenses collect light diffracted by a wafer die aligned with the optic axis and light diffracted by other wafer dies proximally located to such die, rather than light diffracted by the entire wafer. This restriction limits the applicability of the inspection system to dies having many redundant circuit patterns but permits the use of lenses that introduce off-axis aberrations that would otherwise alter the character of the Fourier transform pattern and the filtered defect image.

Such lenses are relatively easy to manufacture because the redundant circuit patterns typically repeat at 50 micron intervals and thereby produce spatial frequency components spaced apart by a distance of about 1.0 millimeter, which is resolvable by conventional optical components. The Fourier transform and imaging areas are preferably of sufficient sizes to accommodate light from only the wafer die aligned with the optic axis. The spatial filter blocks the spatial frequencies of the error-free Fourier transform of such die, i.e., the spatial filter contains only intra-die interference pattern information.

The wafer is positioned in the front focal plane of the Fourier transform lens, and the patterned surface of the wafer is illuminated by a collimated laser beam. The Fourier transform pattern of the illuminated wafer surface is formed in the back focal plane of the Fourier transform lens. A previously fabricated spatial filter is positioned in the plane of the Fourier transform pattern and effectively stops the light transmission from the redundant circuit patterns of the illuminated dies of the wafer but allows the passage of light originating from possible defects.

The inverse Fourier transform lens receives the light either transmitted through or reflected by the spatial filter and performs the inverse Fourier transform on the filtered light diffracted by the illuminated wafer area. Whether the spatial filter is of a type that transmits or reflects light depends on the embodiment of inspection system in which it is incorporated. The filtered image strikes the surface of a two-dimensional photodetector array which detects the presence of light corresponding to defects in only the illuminated on-axis wafer die. The photodetector array is centrally positioned about the optic axis and has a light-sensitive surface area of insufficient size to cover the image plane area in which the defect image corresponding to the on-axis die appears. The inspection of all possible defects in the portions of the wafer surface having many redundant circuit patterns is accomplished by mounting the wafer onto a two-dimensional translation stage and moving the stage so that the illumination area defined by the laser beam continuously scans across the wafer surface from die to die until the desired portions of the wafer surface have been illuminated. The use of a time delay integration technique permits continuous stage movement and inspection of the portions of the wafer surface having many redundant circuit patterns in a stripe-to-stripe raster scan fashion.

The present invention is advantageous because the spatial filter need not be fabricated with the use of an error-free specimen wafer. The reason is that any defects present in such a wafer would produce light of insufficient intensity to expose the spatial filter recording medium.

The present invention detects defects in a specimen pattern by using only intra-die information corresponding to areas of the specimen pattern having many redundant circuit patterns. The premises underlying the inspection method of the present invention are that inter-die interference pattern information is unnecessary if only areas of many redundant patterns are inspected and that inspection of only such areas provides sufficient statistical sampling to determine the defect distribution for the entire specimen pattern.

Additional objects and advantages of the present invention will be apparent from the following detailed description of preferred embodiments thereof, which proceeds with reference to the accompanying drawings.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 is a schematic diagram of the optical components of a first preferred embodiment of the defect inspection system of the present invention.

FIG. 2 is a diagram of a semiconductor wafer comprising a regular array of normally identical dies of the type suitable for defect inspection by the systems of FIGS. 1 and 6.

FIGS. 3A-3C are photographs of an exemplary single die of the semiconductor wafer of FIG. 2 showing within such die a highly redundant circuit pattern for consecutively increasing magnifications.

FIG. 4 is a simplified diagram showing the asymmetry of the Fourier transform and inverse Fourier transform lens system incorporated in the defect inspection system of FIG. 1.

FIG. 5 is a diagram showing the optical elements of the lens system of FIG. 4.

FIG. 6 is a schematic diagram of the optical components of a second preferred embodiment of the defect inspection system of the present invention.

FIG. 7 is a cross sectional view of the spatial filter employed in the defect inspection system of FIG. 6.

FIG. 8 shows the optical components of the Fourier transform and the inverse Fourier transform lens system incorporated in the defect inspection system of FIG. 6.

FIG. 9 is an isometric view of the scanning mechanism for detecting the presence and locations of defects in the semiconductor wafer of FIG. 2.

FIG. 10A is an enlarged fragmentary view showing three stripe regions in the lower left-hand corner of the semiconductor wafer of FIG. 9.

FIG. 10B is an enlarged, not-to-scale view of the stripe regions of FIGS. 9 and 10A that shows the raster scan path followed by the scanning mechanism of FIG. 9 relative to a light sensitive detector to detect defect images in a defect image field.

FIG. 11 is a diagram showing an array of pixel elements in the defect image field under tenfold magnification and an array of light detecting elements of a charge-coupled device used in the present invention.

DETAILED DESCRIPTION OF PREFERRED EMBODIMENTS

FIG. 1 is a schematic diagram of a first preferred embodiment of an inspection system 10 of the present invention that is designed to detect semiconductor wafer defects having a diameter of about one-quarter micron or larger in the presence of a periodic structure comprising many redundant circuit patterns. FIG. 2 is a diagram of a semiconductor wafer 12 of the type inspection system 10 is designed to inspect for defects. Wafer 12 includes a regular array of normally identical dies 14 of which each has at least about twenty redundant circuit patterns 16 along each of the X-axis 18 and Y-axis 20. Each die 14 is typically of square shape with about 3 millimeter sides. FIGS. 3A-3C are photographs of an exemplary single die 14 showing highly repetitive circuit pattern within such element for consecutively increasing magnifications. Although they are of rectangular shape as shown in FIGS. 3A-3C, circuit patterns 16 are assumed for purposes of simplifying the following discussion to be of square shape with about 50 micron sides.

With reference to FIG. 1, inspection system 10 includes a laser source 22 that provides a nearly collimated beam of 442.5 nanometer monochromatic light rays 24 that strike a lens 26 that converges the light rays to a point 28 located in the back focal plane of lens 26. The light rays 30 diverging from focal point 28 strike a small mirror 32 that is positioned a short distance from focal point 28 to reflect a relatively narrow circular beam of light toward a Fourier transform lens section 34, which is shown in FIG. 1 as a single element but which is implemented in five lens elements as will be further described below. Mirror 32 obscures a small region in the center of the Fourier transform plane defined by lens section 34. The size of the obscured region is sufficiently small so that defect information, which is located everywhere in the Fourier transform plane, is only insignificantly blocked by mirror 32.

The effective center of Fourier transform lens section 34 is positioned a distance of slightly less than one focal length away from mirror 32 to provide collimated light rays 36 that strike the patterned surface of wafer 12. Wafer 12 is mounted in a chuck 38 that constitutes part of a two-dimensional translation stage 40. Wafer 12 is positioned in the object or front focal plane 42 of lens section 34, and the collimated light rays 36 illuminate the patterned surface of wafer 12. The collimated light rays 36 illuminate a 20 millimeter diameter area of the surface of wafer 12. The light rays 44 diffracted by the illuminated area of wafer 12 pass through lens section 34 and form the Fourier transform pattern of the illuminated wafer surface in the back focal plane 46 of lens section 34.

The Fourier transform pattern comprises an array of bright spots of light that are distributed in back focal plane 46 in a predictable manner. The 20 millimeter diameter illuminated area of wafer 12 provides a Fourier transform pattern of sufficient accuracy because it is formed from many redundant circuit patterns. The design of lens section 34 is, however, such that it has only a 3 millimeter object field diameter to form in the image plane 60 an essentially aberration-free image of defects in the semiconductor wafer. An entire die can be inspected for defects because translation stage 40 moves the die through the illuminated area. Therefore, a relatively large area of wafer 12 is illuminated to develop an accurate Fourier transform pattern of the redundant circuit patterns, but a lens of relatively small object field diameter collects the light diffracted by the illuminated area to minimize the introduction of aberrations into the Fourier transform pattern as it is formed.

A previously fabricated spatial filter 50 is positioned in the plane 46 of the Fourier transform pattern. Spatial filter 50 can be fabricated in situ by exposing a recording medium, such as a photographic plate, to light diffracted by all of the dies 14 of wafer 12. This can be accomplished with nonerror-free wafer 12 because the defect information carried by light of relatively low intensity would not expose the photographic plate while Fourier transform information carried by relatively high intensity light exposes the photographic plate. Spatial filter 50 can also be fabricated in accordance with known computer generation techniques.

Spatial filter 50 blocks the spatial frequencies of the error-free Fourier transform of the illuminated dies 14 of wafer 12 but allows the passage of light originating from possible defects in, and light diffracted by other wafer dies proximally located to, such dies. The defect-carrying light rays 52 not blocked by spatial filter 50 strike an inverse Fourier transform lens section 54, which is shown schematically as a single lens but includes four lens elements as will be further described below. Inverse Fourier transform lens section 54 performs the inverse Fourier transform on the filtered light pattern of the illuminated wafer dies 14. Lens section 54 is positioned a distance of one focal length away from back focal plane 46 of lens section 34. The elements of lens sections 34 and 54 are aligned along the same optic axis 48, and translation stage 40 moves the wafer dies 14 across the optic axis 48.

A photodetector array 58 is centrally positioned about optic axis 48 in an image plane 60 and receives the image of the defects present in the on-axis portion wafer die 14. Image plane 60 is located in the back focal plane of lens section 54. The magnification of lens section 54 is of an amount that approximately matches the resolution limit of the image to the pixel size of photodetector array 58. In particular, photodetector array 58 has a light sensitive surface 62 whose dimensions are about 10 millimeters.times.10 millimeters within the 30 millimeter diameter image area. A tenfold magnification is, therefore, the proper amount to detect defects in the 3 millimeter diameter area of the on-axis wafer die 14.

To inspect the entire patterned surface of wafer 12, translation stage 40 sequentially moves each portion of the die 14 of wafer 12 to optic axis 48 for illumination by the light emanating from the light source 22. The area of light sensitive surface 62 of the stationary photodetector array 58 limits the amount of light detected to that of a portion of the image corresponding to only the wafer die 14 centered about optic axis 48. The image information corresponding to any portion of illuminated off-axis wafer dies 14 cannot, therefore, reach photodetector array 58. The movement of translation stage 40 is continuous in a stripe-to-stripe raster fashion to implement a time delay integration technique for collecting the defect image information for each die 14 on the patterned surface of wafer 12.

The Fourier transform lens section 34 and inverse Fourier transform lens section 54 are designed as part of one optical system 68 and collectively have ten elements as shown in FIG. 4. The design of optical system 68 is complicated by the stringent requirements for two important design parameters, namely, the minimum spot diameter "d.sub.1 " in Fourier transform plane 46 and minimum spot diameter "d.sub.2 " in image plane 60. A small minimum spot diameter in Fourier transform plane 46 is required to resolve the bright spots produced in such plane by circuit patterns 16 of the largest expected size. If pattern 16 is of square shape, the required spot diameter d.sub.1 satisfies the following expression:

d.sub.1 <<.lambda.f.sub.1 /c,

where .lambda. is the wavelength of light emanating from laser source 22, f.sub.1 is the effective focal length of lens section 34, and c is the length of a side of the square pattern 16. A spot diameter d.sub.1 of 20 microns can be realized for c<300 microns.

A small minimum spot diameter in image plane 60 determines the smallest possible detectable defect size. The minimum spot diameter d.sub.2 is determined by the cooperation of lens sections 34 and 54, and the image magnification "m." A defect of a diameter greater than d.sub.2 /m can be measured from the spatial spread of its image. A defect of a diameter less than d.sub.2 /m, i.e., a subresolution defect, has an image spread that equals d.sub.2 but has an image intensity that decreases quadradically with increasing diameter. To detect subresolution defects, inspection system 10 must be designed to achieve substantially lower electronic or optical noise. The design parameter is achievable with the preferred embodiments of inspection system 10 for d.sub.2 =10 microns, m=10, and d.sub.2 /m=1 micron over the 30 millimeter diameter image field.

With reference to FIGS. 4 and 5, optical system 68 is a near diffraction-limited optical system that accepts light diffracted into a .+-.15.degree.-20.degree. telecentric cone, forms the periodic structure of the Fourier transform of the object (i.e., the redundant circuit patterns of a wafer die 14), and produces a detectable image of one-quarter micron or larger diameter defects. The design of lens section 34 is of asymmetric character to form nearly diffraction-limited light pattern at the Fourier transform plane 46. The reason for the asymmetry is that the diffraction angle of optical system 68 is relatively large (.+-.15.degree.-20.degree.) and the 3 millimeter diameter of the surface to be imaged is moderately large. Lens section 54 requires a relatively long focal length f.sub.2 to achieve the 10X image magnification. The design of lens section 54 is of asymmetric character and the entrance pupil is positioned closer to the front lens element 72 of lens section 54 to balance residual aberrations introduced by lens section 34 and to minimize the length of optical system 68 and the diameters of the lens elements incorporated in it.

The plane of lens section 54 is positioned nearly in contact with Fourier transform plane 46 to provide a compact spatial filtering arrangement. Sufficient space is required between lens section 54 and Fourier transform plane 46 to introduce the illuminating beam through the Fourier transform plane, and to accommodate the mechanical structure that supports spatial filter 50. Optical system 68 is designed so that the image will be an inverted copy of the object with a magnification of -f.sub.2 /f.sub.1, where f.sub.2 =600 millimeters, which is the effective focal length of lens section 54, and f.sub.1 =60 millimeters, which is the effective focal length of lens section 34. Therefore, the magnification "m" equals 10.

Lens section 34 is designed to meet the following two performance requirements. The first and most demanding requirement is that light diffracted into a .+-.15.degree.-20.degree. telecentric cone from any point on a 3 millimeter diameter object placed at the object or front focal plane 42 is collimated with sufficiently small light ray aberrations to permit the ultimate formation of a near diffraction-limited image with very little geometric distortion. The second requirement is that plane waves propagating through a 20 millimeter object diameter over a range of .+-.15.degree.-20.degree. have minimum vignetting and produce a light pattern at Fourier transform plane 46 of less than 20 micron resolution spot diameter.

Residual aberrations introduced by lens section 34 into lens section 54 are magnified and are, therefore, nearly impossible to eliminate by compensating aberrations in lens section 54. The design requirements are, therefore, that lens section 34 at Fourier transform plane 46 be 1) isoplanatic (i.e., the aberrations remain constant over a small section of the Fourier transform image field so that the lens is a linear, shift-invariant filter of spatial frequencies), (2) essentially aplanatic (i.e., free from spherical and coma aberrations), and (3) essentially anastigmatic (i.e., having a flat field with no anastigmatism) for incident plane waves over a .+-.15.degree.-20.degree. diffraction angle and over a 20 millimeter entrance pupil diameter.

To produce a nearly diffraction-limited image at image plane 60, lens section 34 must also produce a plane wave for a point object placed within a 3 millimeter diameter region in the object or front focal plane 42 of the lens. This requires that the chief rays over the .+-.15.degree.-20.degree. diffraction angle range must be forced to be telecentric (i.e., parallel to optic axis 48) during the design of lens section 34 so that residual aberrations presented to lens section 54 will be very small and compensatible.

The design approach of lens section 34 assumes that it receives light emanating from an infinitely distant object with a .+-.15.degree.-20.degree. subtense and an entrance pupil placed at front focal plane 42. The Fourier transform light pattern is, therefore, located at the back focal plane 46 of the lens.

In particular, lens section 34 includes five elements positioned along and centered about optic axis 48. Element 72 is a double-convex lens and element 74 is a positive meniscus lens that are positioned close to the entrance pupil of lens system 68 to control spherical aberrations. A double-concave lens 76 controls the field curvature, and double-convex lens 78 and positive meniscus lens 80 are positioned in the converging beam to control astigmatism. To achieve the aberration control driven by the .+-.15.degree.-20.degree. range of diffraction angles, lens section 34 requires the five elements 72, 74, 76, 78, and 80, which are constructed from glass of a high refractive index.

The design of lens section 54 balances the residual spherical and coma aberrations introduced by lens section 34 when object points are placed at its front focal plane 46 within a 3 millimeter object diameter. A double-convex lens element 70 and a double-concave element 84 are positioned close to Fourier transform plane 46 to cancel the residual spherical aberrations introduced by lens section 34. A negative meniscus lens element 86 cancels the coma aberrations introduced by lens section 34. A positive meniscus lens element 88 of weak positive power distributes the refractive power of lens section 54 so that elements 70, 84, and 86 can introduce the right amount of spherical and coma aberrations to