Electricity: measuring and testing – Fault detecting in electric circuits and of electric components – Of individual circuit component or element
Reexamination Certificate
2000-11-30
2003-05-13
Hoff, Marc S. (Department: 2858)
Electricity: measuring and testing
Fault detecting in electric circuits and of electric components
Of individual circuit component or element
C324S1540PB, C324S754090, C324S758010
Reexamination Certificate
active
06563324
ABSTRACT:
FIELD OF THE INVENTION
The present invention relates to machine vision and more particularly to a machine vision method and apparatus for inspecting semiconductor die assemblies.
BACKGROUND OF THE INVENTION
Integrated circuits are typically assembled using automatic assembly equipment to pick and place a semiconductor die onto a lead frame component where it is usually secured by an adhesive such as epoxy. This process is called die bonding. The lead frame component includes conductive leads that provide external connections to the die. Automatic equipment bonds conductive wires between pads on the die and leads on the lead frame. Precise wire-bonding operations rapidly connect a large number of very small and closely spaced pads to respective leads on the lead frame. Once a wire bonding operation is complete, the die and connective wires are typically encapsulated or enclosed in an insulating package.
Semiconductor die assembly components and manufacturing operations are very expensive so it is important to inspect the surfaces of the semiconductor die after die bonding. This process is called post bond inspection. Deposits of unwanted adhesive on the die are among the most commonly occurring die bonding defects. Such adhesive deposits can effectively “short circuit” the semiconductor die's electronic functions because the adhesive is typically conductive. Electrical testing of a final assembly would detect such short circuits but only after significant additional manufacturing costs have been incurred.
The inspection of semiconductor dies for adhesive has proven to be a vexing machine vision problem. This is a result of the complexity of the background, i.e., the circuitry pattern etched into the layers of the die. Furthermore a die may be rotated and/or shifted by a significant amount when the die and adhesive are placed onto the lead frame, due to lack of fixturing. Such rotation or shifting complicates inspection by a machine vision system because the machine vision system must first find the die.
A semiconductor die is not generally fixtured when it is assembled to a lead frame so it must be first located by any machine vision inspection system prior to its inspection. Machine vision inspection systems must typically perform an alignment procedure to compensate for shifting and rotation prior to performing an inspection operation. Alignment procedures according to the prior art, like the ones using normalized correlation, are configured to only find translation (2 D.O.F.). To find rotation, two models are trained, typically as far apart as possible, to increase accuracy. By determining the two positions, the rotation component can be computed. However, if the rotation is excessive the position may not be found. This procedure involves training a model in a reference image and then finding it within a region of interest in the runtime image. Fiducials may be provided on the reference object to increase the accuracy and efficiency of the alignment procedure. An alignment procedure that is used in semiconductor die inspection typically recognizes and locates two opposite corners of a die under inspection and two opposite corners of the lead frame. The machine vision inspection system can then compute the shift and rotation of the semiconductor die relative to its train-time position.
Methods according to the prior art have limited tolerance to rotation. For example certain of such methods can tolerate up to 10 degrees or 15 degrees of rotation but at the cost of reduced accuracy. Alternatively, numerous incrementally rotated reference images could be stored, but such methods are inefficient because they consume excessive memory and processor resources. Such methods decrease system accuracy and are inefficient because they consume excessive memory and processor resources.
The prior art suggests the use of a technique referred to as golden template comparison (GTC) to inspect die surfaces. GTC is a technique for locating objects by comparing a feature under scrutiny (to wit, a die surface) to a good image—or golden template—that is stored in memory. The technique subtracts the good image from the test image and analyzes the difference to determine if the unexpected object (e.g., a defect) is present. For example, upon subtracting the image of a good die surface from a defective one, the resulting “difference” image would reveal an adhesive blotch that could be flagged as a defect.
Before GTC inspections can be performed, the system must be “trained” so that the golden template can be stored in memory. To this end, the GTC training functions are employed to analyze several good samples of a scene to create a “mean” image and “standard deviation” image. The mean image is a statistical average of all the samples analyzed by the training functions. It defines what a typical good scene looks like. The standard deviation image defines those areas on the object where there is little variation from part to part, as well as those areas in which there is great variation from part to part. This latter image permits GTC's runtime inspection functions to use less sensitivity in areas of greater expected variation, and more sensitivity in areas of less expected variation.
At runtime, a system employing GTC captures an image of a scene of interest. Where the position of that scene is different from the training position, the captured image is aligned, or registered, with the mean image. The intensities of the captured image are also normalized with those of the mean image to ensure that the variations in illumination do not adversely affect the comparison.
The GTC inspection functions then subtract the registered, normalized, captured image that contains all the variations between the two. That difference image is then compared with at “threshold” image derived from the standard deviation image. This determines which pixels of the difference image are to be ignored and which should be analyzed as possible defects. The latter are subjected to morphology, to eliminate or accentuate pixel data patterns and to eliminate noise. An object recognition technique, such as connectivity analysis, can then be employed to classify the apparent defects.
Although GTC inspection tools have proven quite successful, they suffer some limitations. For example, except in unusual circumstances, GTC requires registration—i.e., that the image under inspection be registered with the template image. GTC also used a standard deviation image for thresholding, which can result in a loss of resolution near edges due to high resulting threshold values. GTC is, additionally, limited to applications where the images are repeatable: it cannot be used where image-to-image variation results from changes in size, shape, orientation and warping. Furthermore GTC methods disadvantageously tolerate rotation only by inefficiently storing a set of rotated referenced images as previously described.
In application to die surface inspection, GTC is further limited because its fixed template typically does not include the die edges because edges are not generally repeatable with the necessary precision from die-to die due to manufacturing machinery sawing tolerances. It is here, however, that the probability of deposited adhesive is very high. Moreover, the complexity of the etching patterns on the die surfaces results in large area being effectively masked by the high standard deviation. Therefore, GTC methods are generally incapable of inspecting the fine detail of typical semiconductor die surfaces.
An improved method of inspecting semiconductor dies is taught in U.S. Pat. No. 5,949,901 which is incorporated herein by reference in its entirety. The method according to U.S. Pat. No. 5,949,901 (the '901 invention) includes the steps of generating a first image of the die (including, the patterns etched into its surface and any other structures—together, referred to as the “die,” or “die surface” or “background”), generating a second image of the die and any defects thereon, and subtracting the second image from the firs
Cognex Technology and Investment Corporation
Hoff Marc S.
Michaelis Brian
Sundaram T. R.
LandOfFree
Semiconductor device image inspection utilizing rotation... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Semiconductor device image inspection utilizing rotation..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Semiconductor device image inspection utilizing rotation... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3033150