Issue



Inspecting Post Wire-bond Interconnects


04/01/2006







An AOI Approach

BY GEORGE T. AYOUB, MACHINE VISION PRODUCTS INC.

Wire bond technology will continue to prosper in many sectors of the electronics packaging industry well into the foreseeable future. Major trends in this industry over the years have included a continuous increase in the number of interconnections, circuit miniaturization, industry emphasis on speed of assembly, and cost reduction per interconnection. Wire-bond machines have kept up with these trends, and are sophisticated, reliable, fast, and accurate. However, wire-bond inspection lacks the means to automate inspection and ensure the integrity of wire-bond interconnections, which directly impact the quality of the end product. As the number of interconnections increases, the opportunity to produce a defective component multiplies. Because wire bonding takes place at the end of production, the cost of a bad interconnection is high relative to a defect that can be detected and corrected at the beginning of the process. Therefore, a bad interconnection is a risk that impacts cost and quality of the product.

Currently, most inspection methods for wire bond are manual, and use visual check with a microscope, contact inspection, or semi-automated inspection assisted by an optical or X-ray imaging sensor. These inspection methods are slow, labor-intensive, and costly. Because of these limitations, they are often used to test the product on a sample basis. Manual methods - both visual and those assisted by a sensor - are far from perfect and suffer from the inherent variability of human inspection. Due to the lack of automated measurements, they are subjective and dependent on the operator. Contact inspection tests the security of the wire bond by means of physical contact. This method is slow and risks physical damage due to contact or potential electrostatic damage. All these methods are limited to wire-bond inspection, which is another drawback. Inspection tool capabilities should encompass measurement of die placement and inspect for solder joint quality of other components in the vicinity of the wire bonds. There is an urgent need today for an efficient and reliable method of inspection that is effective, safe, dependable, measurement-driven, capable of inspecting all wire-bond failure modes, versatile enough to include measurement and inspection of other electronic components, and fast enough to keep up with the production while inspecting 100% of the products.

Parallel to development in the packaging industry, rule-based automated optical inspection (AOI) has emerged as an effective inspection and measurement method for all process steps in PCB assembly. AOI has become a proven, reliable tool for inspecting solder paste, component placement, and solder joint inspection, and has been widely used to improve quality and reduce assembly costs. Notable advances in AOI have been driven by advances in camera technology and by the availability of fast and economical computing platforms. Today’s AOI uses fast, sensitive camera sensors and a multitude of programmable LED illuminators powered by sophisticated algorithms for inspection and measurements, and are able to meet requirements for 100% inspection at production line speed. This has resulted in increased defect coverage, higher inspection speed, and lower false accepts and rejects. The programming aspects of this technology have also become easier over the years, contributing significantly to its widespread use. The question naturally arises: can AOI technology provide a basis to meet the stiff demands for post-wire-bond inspection?

The answer to this question is “yes.” In the past, the AOI industry fell short of meeting the requirements presented by post-wire-bond inspection. Until recently, there was no universal equipment capable of meeting all these requirements. Now, a tool has been introduced that provides a useful solution for many aspects of post-wire-bond inspection.

The ability to extract the wire from the complex varying background between die and pad is an important aspect of post-wire-bond inspection. Accomplishing this task requires smart illumination and inspection algorithms to work together, increasing the signal-to-noise ratio between the wire and its surroundings. The inspection tool uses a large color camera sensor and custom-built, programmable colored LED illuminators at different angles, with respect to the wire bond. The wire’s metallic surface reflects the light and can appear black or white with respect to the background, depending on the height of the illuminator angle. The key to increasing the signal-to-noise ratio is to use all the angles of light to better extract the wire from the background. This task is accomplished with proprietary sophisticated algorithms that work hand-in-hand with the illuminator. The algorithm checks if the wire is registered in the right region on the die and pad. Then, it assesses the quality of the connection with the die and pad, and checks for scratches in these regions. Next, the wire is traced and examined for continuity, straightness, and maximum deviation from a straight-fitted line. The loop height is checked to verify that it conforms to a given tolerance by the wire reflection at different lighting illumination angles. All the algorithms use digital filters in a sequential manner to extract features and examine signatures using measurements at each step.

Die translation and rotation with respect to its ideal position is measured in sub-pixel accuracy using many windows around the edges to minimize errors. Registration of the die relies on the stage accuracy, as well as accurate fiducial and CAD information. The tool is also able to measure the position of other components in the circuit, and ascertain the quality of the solder joints, flagging any defects.

The large format sensor and the proprietary frame grabber allows images to be captured “on the fly” while the camera is moving, meeting resolution and speed requirements. The field-of-view is small to ensure an adequate number of pixels on target. Moreover, the system is equipped with an illuminator to ensure image quality and depth of focus. The illumination and algorithm approach is the same for thin and thick wires; however, the resolution of the camera, measured in µm/pixel, is different in each case to optimize speed of inspection (Figures 1 and 2).


Figures 1 and 2. Both thin (Figure 1, above) and thick (Figure 2, below) wires can be inspected with a simple change in camera resolution. Because the camera resolution is measured in µm/pixel, optimizing the overall inspection speed is different in each case.
Click here to enlarge image

null

Click here to enlarge image

The tool goes beyond the detection of pass/fail defects, and assists in enhancing yield through statistical process control (SPC) techniques on both attribute and measured variables. The SPC package is an integral part of the tool and tracks any measurement in real time, allowing the operator to take control actions if limits exceed normal expectancy. Preventing defects is critical in keeping the process under control. Depending on the alarm setting, the system is able to stop the line and turn on a yellow or red light for visual feedback to the operator.


Figure 3. A high-precision AOI machine.
Click here to enlarge image

Preliminary results show the tool’s capability to trace wires with thicknesses varying between 0.5 to 10 mils, with success against complex backgrounds. The numbers reported for measurement accuracy and repeatability show that die translation can be measured accurately to <10 µm at three standard deviations, and its rotation to <0.05°.

This innovation is a first attempt to meet the challenge of post-wire-bond inspection. Future work will continue to enhance the signal to noise ratio, to extend the defect coverage for post-wire-bond to multiple layers, and to increase the speed of inspection.

GEORGE T. AYOUB, Ph.D., president and CEO, may be contacted at Machine Vision Products Inc., 5940 Darwin Ct., Carlsbad, CA 92008; 760/438-1138; E-mail: george@visionpro.com.