Electricity: measuring and testing – Fault detecting in electric circuits and of electric components – Of individual circuit component or element
Reexamination Certificate
1999-09-14
2001-05-08
Metjahic, Safet (Department: 2858)
Electricity: measuring and testing
Fault detecting in electric circuits and of electric components
Of individual circuit component or element
C324S1540PB
Reexamination Certificate
active
06229331
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to an apparatus for, and a method of, inspecting patterns on semiconductor integrated devices and, particularly, to an apparatus for, and a method of, optically or electro-optically inspecting the surfaces of semiconductor wafers and photomasks for defects.
2. Description of the Related Art
An apparatus according to a prior art for inspecting the appearance of semiconductor integrated devices photographs two adjacent semiconductor devices or dies, compares the images of the dies pixel by pixel, and determines that one of the dies is defective if the images disagree with each other. This prior art employs a photographing apparatus such as a combination of an optical microscope and TDI (time delay integration) image pickup elements. The photographing apparatus continuously scans a first die, which may be a defective die, in a row direction, obtains a multivalued image of the first die, and stores the image in an image memory. Similarly, the photographing apparatus picks up a multivalued image of a second die, which is a reference die and is arranged adjacent to the first die, and stores the image in the image memory. The apparatus reads the images frame by frame and compares the gray levels of corresponding pixels in the frames. Any pixel that provides a gray level difference above a threshold with respect to a corresponding pixel of the reference image is determined to be defective. This technique usually provides satisfactory defect detecting sensitivity in appearance inspection.
Recently, however, the requirements for the defect detecting sensitivity of appearance inspection have increasing due to finer design rules for semiconductor devices. To improve the defect detecting sensitivity of an inspection apparatus, semiconductor devices to be inspected must exhibit less color unevenness or process noise such as metal grain caused by surface irregularities such as metal wiring. In practice, however, it is difficult to eliminate such process noise from semiconductor devices. A location on a semiconductor device where a lot of metal grain appears scatters light to reduce light quantity entering into an object lens of a microscope, thereby lowering the gray level of a corresponding location on an image picked up from the semiconductor device. The process noise is not a critical defect such as a short circuit between wires, and therefore, it is preferable not detected. If a fixed threshold is applied to an area where the process noise is present when checking a difference between the gray levels of two pixels, the process noise will be recognized as a defect, thereby lowering the defect detecting sensitivity.
To solve this problem, experiments have been made to set higher thresholds for brighter areas where metal wires are present and lower thresholds for dark areas between the metal wires, so that process noise such as metal grain in the metal wires may not be detected as defects and critical defects such as short circuits between the wires are surely detected. For example, one prior art compares two images with each other. In each of the images, the prior art defines an area of 3×3 pixels and calculates a mean value of the gray levels of the 9 pixels in the area as well as a difference (hereinafter referred to as a range value) between the maximum and minimum of the gray levels of the 9 pixels in the area. According to at least one of the mean value and range value, the prior art selects a threshold for a center pixel of the area to test the difference between the two images. Namely, each pixel in the images is provided with a proper threshold according to a mean value and/or a range value that reflect the gray levels of peripheral pixels, to improve defect detecting sensitivity.
This prior art will be explained in more detail with reference to
FIGS. 15
to
17
.
FIG. 15
shows an image picked up from metal wires that are formed at intervals of 0.35 &mgr;m with each wire having a width of 0.35 &mgr;m. The image is made of pixels each of 0.20 &mgr;m in size. The metal wires
151
to
154
are alternated with spaces
155
to
158
. Numerals
1
to
10
on the left side of the image are row numbers of the pixels, and numerals
1
to
14
on the top of the image are column numbers of the pixels. Each pixel is represented as (a, b) where a is a row number and b is a column number. For the sake of simplicity of explanation, the metal wires
151
to
154
involve no metal grain. The columns
3
,
7
,
10
, and
14
are entirely in the metal wires
151
to
154
, and therefore, are brightest to provide a gray level of
225
. The columns
1
,
5
,
8
, and
12
are entirely in the spaces
155
to
158
, and therefore, are darkest to provide a gray level of
25
. The columns
2
,
4
,
6
,
9
,
11
, and
13
involve a mixture of metal wire and space to provide a gray level that is dependent on the ratio of the metal wire and space. For example, pixels in the column
6
have each a gray level of
175
, pixels in the column
9
a gray level of
75
, and pixels in the column
11
a gray level of
125
. In practice, the gray levels of pixels in a given column are not equal to one another and vary from one to another due to process noise such as metal grain in wiring areas and boundaries between bright and dark areas. Such boundaries scatter light to further drop gray levels.
The prior art calculates at least a mean value or a range value of the gray levels of every 3×3 pixels in a given image and, according to the mean or range values calculated, divides pixels of the image into groups. Thereafter, the prior art sets a threshold for each group.
FIG. 16
shows mean values calculated for the pixels of the image of FIG.
15
. The mean value of each pixel is calculated from the gray levels of 3×3 pixels with the pixel in question being at the center of the 3×3 pixels. For example, a pixel (
2
,
2
) has a mean value of 108, which is calculated by averaging the gray levels of 3×3 pixels around the pixel (
2
,
2
), i.e., (25×3+75×3+225×3)/9=108.3. Mean values for the other pixels are calculated similarly. Instead of directly employing the gray levels of pixels to find thresholds, this prior art employs the mean gray levels of pixels to find thresholds that cope with local conditions such as noise, darkness, and brightness, thereby improving defect detecting sensitivity.
FIG. 17
shows range values calculated for the pixels of the image of FIG.
15
. The range value of each pixel is calculated from the difference between the maximum and minimum of the gray levels of 3×3 pixels with the pixel in question being at the center of the 3×3 pixels. For example, the pixel (
2
,
2
) has a range value of 200, which is calculated from the difference between a maximum of 225 and a minimum of 25 among the gray levels of 3×3 pixels around the pixel (
2
,
2
). Range values for the other pixels are calculated similarly. Instead of directly employing the gray levels of pixels to find thresholds, this prior art employs the range values of pixels to find thresholds that cope with local conditions such as noise, darkness, and brightness, thereby improving defect detecting sensitivity.
Depending on the nature of an object and defects to be detected, the mean values of FIG.
16
and the range values of
FIG. 17
are employed to improve defect detecting sensitivity irrespective of the presence of process noise that may fluctuate the gray levels of pixels.
It is essential for the appearance inspection apparatuses to surely detect critical defects, i.e., short circuits between metal wires. To detect such critical defects, spaces between metal wires must surely be recognized. Once the spaces are recognized, proper thresholds are set for the spaces to surely detect defects in the spaces.
The prior art of
FIGS. 15
to
17
is imperfect to distinguish spaces from wiring areas, and therefore, is unable to completely detect critical defects such as short circui
Christie Parker & Hale LLP
Metjahic Safet
Tang Minh
Tokyo Seimitsu Co. Ltd.
LandOfFree
Apparatus for and method of inspecting patterns on... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Apparatus for and method of inspecting patterns on..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Apparatus for and method of inspecting patterns on... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2461366